public inbox for [email protected]help / color / mirror / Atom feed
[PATCH v9 1/6] snapshot scalability: Don't compute global horizons when building snapshots. 29+ messages / 4 participants [nested] [flat]
* [PATCH v9 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-04-08 11:33 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-04-08 11:33 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 11 +- src/include/access/transam.h | 100 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 39 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 19 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 +++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 6 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 92 +- src/backend/access/transam/varsup.c | 50 + src/backend/access/transam/xlog.c | 11 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 1022 +++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 ++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 39 files changed, 1630 insertions(+), 611 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index f279edc4734..ef2fcb55a71 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -201,11 +204,15 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +struct GlobalVisState; +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 9a808f64ebe..94ba797f026 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -54,6 +54,8 @@ #define FullTransactionIdFollowsOrEquals(a, b) ((a).value >= (b).value) #define FullTransactionIdIsValid(x) TransactionIdIsValid(XidFromFullTransactionId(x)) #define InvalidFullTransactionId FullTransactionIdFromEpochAndXid(0, InvalidTransactionId) +#define FirstNormalFullTransactionId FullTransactionIdFromEpochAndXid(0, FirstNormalTransactionId) +#define FullTransactionIdIsNormal(x) FullTransactionIdFollowsOrEquals(x, FirstNormalFullTransactionId) /* * A 64 bit value that contains an epoch and a TransactionId. This is @@ -93,15 +95,48 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ static inline void FullTransactionIdAdvance(FullTransactionId *dest) { dest->value++; + + /* + * In contrast to 32bit XIDs don't step over the "actual" special xids. + * For 64bit xids these can't be reached as part of a wraparound as they + * can in the 32bit case. + */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + /* + * But we do need to step over XIDs that'd appear special only for 32bit + * XIDs. + */ while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) dest->value++; } +/* + * Retreat a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdRetreat(FullTransactionId *dest) +{ + dest->value--; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value--; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -193,8 +228,8 @@ typedef struct VariableCacheData /* * These fields are protected by ProcArrayLock. */ - TransactionId latestCompletedXid; /* newest XID that has committed or - * aborted */ + FullTransactionId latestCompletedFullXid; /* newest full XID that has + * committed or aborted */ /* * These fields are protected by CLogTruncationLock @@ -244,6 +279,12 @@ extern void AdvanceOldestClogXid(TransactionId oldest_datfrozenxid); extern bool ForceTransactionIdLimitUpdate(void); extern Oid GetNewObjectId(void); +#ifdef USE_ASSERT_CHECKING +extern void AssertTransactionInAllowableRange(TransactionId xid); +#else +#define AssertTransactionInAllowableRange(xid) ((void)true) +#endif + /* * Some frontend programs include this header. For compilers that emit static * inline functions even when they're unused, that leads to unsatisfied @@ -260,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index ae4f573ab46..23d12c1f72f 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,13 +42,6 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ @@ -56,7 +49,6 @@ struct XidCache #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index a5c7d0c0644..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,41 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_ANALYZE_FLAG 0x04 /* currently running - * analyze */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_ANALYZE_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG -/* Ignore analyze backends */ -#define PROCARRAY_FLAGS_ANALYZE PROCARRAY_FLAGS_DEFAULT | PROCARRAY_ANALYZE_FLAG -/* Ignore both vacuum and analyze backends */ -#define PROCARRAY_FLAGS_VACUUM_ANALYZE PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG | PROCARRAY_ANALYZE_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -88,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index b28d13ce841..3ddc526febc 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index b60dba052fa..af4731cff18 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * master and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on master and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index c4a5aa616a3..d572a0f01d3 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1535,6 +1535,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1545,7 +1546,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1634,9 +1636,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch @@ -2192,8 +2199,8 @@ heap_multi_insert(Relation relation, TupleTableSlot **slots, int ntuples, RelationPutHeapTuple(relation, buffer, heaptuples[ndone], false); /* - * Note that heap_multi_insert is not used for catalog tuples yet, - * but this will cover the gap once that is the case. + * Note that heap_multi_insert is not used for catalog tuples yet, but + * this will cover the gap once that is the case. */ if (needwal && need_cids) log_heap_new_cid(relation, heaptuples[ndone]); diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 56b35622f1a..659fc4d8697 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1201,7 +1201,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1242,6 +1242,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1261,17 +1272,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index dba10890aab..b25b3e429ed 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 1794cfd8d9a..453465c54a9 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index f3382d37a40..22e86a391b4 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -780,6 +780,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -808,6 +809,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize the state for a parallel vacuum. As of now, only one worker * can be used for an index, so we invoke parallelism only if there are at @@ -1231,7 +1234,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1588,14 +1592,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index a3f77169a79..c6276e6fe86 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index 2d0f8f4b79a..42993f131db 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -336,9 +336,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -405,8 +405,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index 39b8f17f4b5..37e8c97b0c9 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -983,7 +983,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2186,8 +2186,8 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, bool *rightsib_empty) * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration - * of that scan. + * will continue to do so, holding back xmin horizon, for the duration of + * that scan. */ page = BufferGetPage(buf); opaque = (BTPageOpaque) PageGetSpecialPointer(page); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 36294789f3f..9afab9e2111 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -802,6 +802,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -811,12 +817,11 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* - * If oldest btpo.xact in the deleted pages is older than - * RecentGlobalXmin, then at least one deleted page can be recycled. + * If oldest btpo.xact in the deleted pages is visible to everyone, + * then at least one deleted page can be recycled. */ result = true; } @@ -1227,14 +1232,13 @@ restart: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index 99d0914e724..c6b6a723dc9 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -926,11 +926,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * master and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on master and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..4e2178dabab 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -257,31 +257,31 @@ simultaneously, we have one backend take ProcArrayLock and clear the XIDs of multiple processes at once.) ProcArrayEndTransaction also holds the lock while advancing the shared -latestCompletedXid variable. This allows GetSnapshotData to use -latestCompletedXid + 1 as xmax for its snapshot: there can be no +latestCompletedFullXid variable. This allows GetSnapshotData to use +latestCompletedFullXid + 1 as xmax for its snapshot: there can be no transaction >= this xid value that the snapshot needs to consider as completed. In short, then, the rule is that no transaction may exit the set of -currently-running transactions between the time we fetch latestCompletedXid +currently-running transactions between the time we fetch latestCompletedFullXid and the time we finish building our snapshot. However, this restriction only applies to transactions that have an XID --- read-only transactions can end without acquiring ProcArrayLock, since they don't affect anyone -else's snapshot nor latestCompletedXid. +else's snapshot nor latestCompletedFullXid. Transaction start, per se, doesn't have any interlocking with these considerations, since we no longer assign an XID immediately at transaction start. But when we do decide to allocate an XID, GetNewTransactionId must store the new XID into the shared ProcArray before releasing XidGenLock. -This ensures that all top-level XIDs <= latestCompletedXid are either +This ensures that all top-level XIDs <= latestCompletedFullXid are either present in the ProcArray, or not running anymore. (This guarantee doesn't apply to subtransaction XIDs, because of the possibility that there's not room for them in the subxid array; instead we guarantee that they are present or the overflow flag is set.) If a backend released XidGenLock before storing its XID into MyPgXact, then it would be possible for another -backend to allocate and commit a later XID, causing latestCompletedXid to +backend to allocate and commit a later XID, causing latestCompletedFullXid to pass the first backend's XID, before that value became visible in the -ProcArray. That would break GetOldestXmin, as discussed below. +ProcArray. That would break ComputeXidHorizons, as discussed below. We allow GetNewTransactionId to store the XID into MyPgXact->xid (or the subxid array) without taking ProcArrayLock. This was once necessary to @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedFullXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/varsup.c b/src/backend/access/transam/varsup.c index 2570e7086a7..c12e477ecfc 100644 --- a/src/backend/access/transam/varsup.c +++ b/src/backend/access/transam/varsup.c @@ -566,3 +566,53 @@ GetNewObjectId(void) return result; } + + +#ifdef USE_ASSERT_CHECKING + +/* + * Assert that xid is between [oldestXid, nextFullXid], which is the range we + * expect XIDs coming from tables etc to be in. + * + * As ShmemVariableCache->oldestXid could change just after this call without + * further precautions, and as a wrapped-around xid could again fall within + * the valid range, this assertion can only detect if something is definitely + * wrong, but not establish correctness. + * + * This intentionally does not expose a return value, to avoid code being + * introduced that depends on the return value. + */ +void +AssertTransactionInAllowableRange(TransactionId xid) +{ + TransactionId oldest_xid; + TransactionId next_xid; + + Assert(TransactionIdIsValid(xid)); + + /* we may see bootstrap / frozen */ + if (!TransactionIdIsNormal(xid)) + return; + + /* + * We can't acquire XidGenLock, as this may be called with XidGenLock + * already held (or with other locks that don't allow XidGenLock to be + * nested). That's ok for our purposes though, since we already rely on + * 32bit reads to be atomic. While nextFullXid is 64 bit, we only look at + * the lower 32bit, so a skewed read doesn't hurt. + * + * There's no increased danger of falling outside [oldest, next] by + * accessing them without a lock. xid needs to have been created with + * GetNewTransactionId() in the originating session, and the locks there + * pair with the memory barrier below. We do however accept xid to be <= + * to next_xid, instead of just <, as xid could be from the procarray, + * before we see the updated nextFullXid value. + */ + pg_memory_barrier(); + oldest_xid = ShmemVariableCache->oldestXid; + next_xid = XidFromFullTransactionId(ShmemVariableCache->nextFullXid); + + Assert(TransactionIdFollowsOrEquals(xid, oldest_xid) || + TransactionIdPrecedesOrEquals(xid, next_xid)); +} +#endif diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index c38bc1412d8..7a7d0dd31ef 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -7837,10 +7837,11 @@ StartupXLOG(void) XLogCtl->lastSegSwitchTime = (pg_time_t) time(NULL); XLogCtl->lastSegSwitchLSN = EndOfLog; - /* also initialize latestCompletedXid, to nextXid - 1 */ + /* also initialize latestCompletedFullXid, to nextFullXid - 1 */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); - ShmemVariableCache->latestCompletedXid = XidFromFullTransactionId(ShmemVariableCache->nextFullXid); - TransactionIdRetreat(ShmemVariableCache->latestCompletedXid); + ShmemVariableCache->latestCompletedFullXid = + ShmemVariableCache->nextFullXid; + FullTransactionIdRetreat(&ShmemVariableCache->latestCompletedFullXid); LWLockRelease(ProcArrayLock); /* @@ -9051,7 +9052,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9411,7 +9412,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index 924ef37c816..34b71b6c1c5 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1056,7 +1056,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 3a89f8fe1e2..77474b8d7d6 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -957,8 +957,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1347,12 +1364,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1683,8 +1701,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1707,8 +1726,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 7e97ffab27d..df1af9354ce 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1878,6 +1878,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index aec885e9871..158b2f3d73b 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d69fb90132d..86336a10da5 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index 122d884f3e4..d8989762d74 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2097,9 +2097,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2112,10 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 363000670b2..58f119f9895 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,143 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* familiy of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedFullXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +255,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -175,6 +328,10 @@ static void KnownAssignedXidsReset(void); static inline void ProcArrayEndTransactionInternal(PGPROC *proc, PGXACT *pgxact, TransactionId latestXid); static void ProcArrayGroupClearXid(PGPROC *proc, TransactionId latestXid); +static void MaintainLatestCompletedXid(TransactionId latestXid); +static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); + +static inline FullTransactionId FullXidViaRelative(FullTransactionId rel, TransactionId xid); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -351,9 +508,7 @@ ProcArrayRemove(PGPROC *proc, TransactionId latestXid) Assert(TransactionIdIsValid(allPgXact[proc->pgprocno].xid)); /* Advance global latestCompletedXid while holding the lock */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, - latestXid)) - ShmemVariableCache->latestCompletedXid = latestXid; + MaintainLatestCompletedXid(latestXid); } else { @@ -466,9 +621,7 @@ ProcArrayEndTransactionInternal(PGPROC *proc, PGXACT *pgxact, pgxact->overflowed = false; /* Also advance global latestCompletedXid while holding the lock */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, - latestXid)) - ShmemVariableCache->latestCompletedXid = latestXid; + MaintainLatestCompletedXid(latestXid); } /* @@ -623,6 +776,58 @@ ProcArrayClearTransaction(PGPROC *proc) pgxact->overflowed = false; } +/* + * Update ShmemVariableCache->latestCompletedFullXid to point to latestXid if + * currently older. + */ +static void +MaintainLatestCompletedXid(TransactionId latestXid) +{ + FullTransactionId cur_latest = ShmemVariableCache->latestCompletedFullXid; + + Assert(LWLockHeldByMe(ProcArrayLock)); + Assert(FullTransactionIdIsValid(cur_latest)); + + if (TransactionIdPrecedes(XidFromFullTransactionId(cur_latest), latestXid)) + { + ShmemVariableCache->latestCompletedFullXid = + FullXidViaRelative(cur_latest, latestXid); + } + + Assert(IsBootstrapProcessingMode() || + FullTransactionIdIsNormal(ShmemVariableCache->latestCompletedFullXid)); +} + +/* + * Same as MaintainLatestCompletedXid, except for use during WAL replay. + */ +static void +MaintainLatestCompletedXidRecovery(TransactionId latestXid) +{ + FullTransactionId cur_latest = ShmemVariableCache->latestCompletedFullXid; + FullTransactionId rel; + + Assert(AmStartupProcess() || !IsUnderPostmaster); + Assert(LWLockHeldByMe(ProcArrayLock)); + + /* + * Need a FullTransactionId to compare latestXid with. Can't rely on + * latestCompletedFullXid to be initialized in recovery. But in recovery + * it's safe to access nextFullXid without a lock for the startup process. + */ + rel = ShmemVariableCache->nextFullXid; + Assert(FullTransactionIdIsValid(ShmemVariableCache->nextFullXid)); + + if (!FullTransactionIdIsValid(cur_latest) || + TransactionIdPrecedes(XidFromFullTransactionId(cur_latest), latestXid)) + { + ShmemVariableCache->latestCompletedFullXid = + FullXidViaRelative(rel, latestXid); + } + + Assert(FullTransactionIdIsNormal(ShmemVariableCache->latestCompletedFullXid)); +} + /* * ProcArrayInitRecovery -- initialize recovery xid mgmt environment * @@ -843,7 +1048,7 @@ ProcArrayApplyRecoveryInfo(RunningTransactions running) * Now we've got the running xids we need to set the global values that * are used to track snapshots as they evolve further. * - * - latestCompletedXid which will be the xmax for snapshots + * - latestCompletedFullXid which will be the xmax for snapshots * - lastOverflowedXid which shows whether snapshots overflow * - nextXid * @@ -869,14 +1074,11 @@ ProcArrayApplyRecoveryInfo(RunningTransactions running) /* * If a transaction wrote a commit record in the gap between taking and - * logging the snapshot then latestCompletedXid may already be higher than - * the value from the snapshot, so check before we use the incoming value. + * logging the snapshot then latestCompletedFullXid may already be higher + * than the value from the snapshot, so check before we use the incoming + * value. It also might not yet be set at all. */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, - running->latestCompletedXid)) - ShmemVariableCache->latestCompletedXid = running->latestCompletedXid; - - Assert(TransactionIdIsNormal(ShmemVariableCache->latestCompletedXid)); + MaintainLatestCompletedXidRecovery(running->latestCompletedXid); LWLockRelease(ProcArrayLock); @@ -1050,10 +1252,11 @@ TransactionIdIsInProgress(TransactionId xid) LWLockAcquire(ProcArrayLock, LW_SHARED); /* - * Now that we have the lock, we can check latestCompletedXid; if the + * Now that we have the lock, we can check latestCompletedFullXid; if the * target Xid is after that, it's surely still running. */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, xid)) + if (TransactionIdPrecedes(XidFromFullTransactionId(ShmemVariableCache->latestCompletedFullXid), + xid)) { LWLockRelease(ProcArrayLock); xc_by_latest_xid_inc(); @@ -1250,159 +1453,183 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See ComputedXidHorizonsResult for the various computed horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedFullXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * in the master but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); - /* - * We initialize the MIN() calculation with latestCompletedXid + 1. This - * is a lower bound for the XIDs that might appear in the ProcArray later, - * and so protects us against overestimating the result due to future - * additions. - */ - result = ShmemVariableCache->latestCompletedXid; - Assert(TransactionIdIsNormal(result)); - TransactionIdAdvance(result); + h->latest_completed = ShmemVariableCache->latestCompletedFullXid; - for (index = 0; index < arrayP->numProcs; index++) + /* + * We initialize the MIN() calculation with latestCompletedFullXid + 1. + * This is a lower bound for the XIDs that might appear in the ProcArray + * later, and so protects us against overestimating the result due to + * future additions. + */ + { + TransactionId initial; + + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1414,34 +1641,143 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1492,12 +1828,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1508,11 +1842,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; + FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1556,13 +1891,16 @@ GetSnapshotData(Snapshot snapshot) */ LWLockAcquire(ProcArrayLock, LW_SHARED); + latest_completed = ShmemVariableCache->latestCompletedFullXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ - xmax = ShmemVariableCache->latestCompletedXid; - Assert(TransactionIdIsNormal(xmax)); + xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); + Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1591,12 +1929,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1712,34 +2044,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidViaRelative(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidViaRelative(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidViaRelative(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -1986,7 +2362,7 @@ GetRunningTransactionData(void) LWLockAcquire(ProcArrayLock, LW_SHARED); LWLockAcquire(XidGenLock, LW_SHARED); - latestCompletedXid = ShmemVariableCache->latestCompletedXid; + latestCompletedXid = XidFromFullTransactionId(ShmemVariableCache->latestCompletedFullXid); oldestRunningXid = XidFromFullTransactionId(ShmemVariableCache->nextFullXid); @@ -3209,9 +3585,7 @@ XidCacheRemoveRunningXids(TransactionId xid, elog(WARNING, "did not find subXID %u in MyProc", xid); /* Also advance global latestCompletedXid while holding the lock */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, - latestXid)) - ShmemVariableCache->latestCompletedXid = latestXid; + MaintainLatestCompletedXid(latestXid); LWLockRelease(ProcArrayLock); } @@ -3238,6 +3612,276 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + GlobalVisSharedRels.maybe_needed = + FullXidViaRelative(horizons.latest_completed, + horizons.shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidViaRelative(horizons.latest_completed, + horizons.catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidViaRelative(horizons.latest_completed, + horizons.data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidViaRelative(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + +/* + * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it + * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). + * + * Be very careful about when to use this function. It can only safely be used + * when there is a guarantee that xid is within MaxTransactionId / 2 xids of + * rel. That e.g. can be guaranteed if the the caller assures a snapshot is + * held by the backend and xid is from a table (where vacuum/freezing ensures + * the xid has to be within that range), or if xid is from the procarray and + * prevents xid wraparound that way. + */ +static inline FullTransactionId +FullXidViaRelative(FullTransactionId rel, TransactionId xid) +{ + TransactionId rel_xid = XidFromFullTransactionId(rel); + + Assert(TransactionIdIsValid(xid)); + Assert(TransactionIdIsValid(rel_xid)); + + /* not guaranteed to find issues, but likely to catch mistakes */ + AssertTransactionInAllowableRange(xid); + + return FullTransactionIdFromU64(U64FromFullTransactionId(rel) + + (int32) (xid - rel_xid)); +} + /* ---------------------------------------------- * KnownAssignedTransactionIds sub-module @@ -3390,9 +4034,7 @@ ExpireTreeKnownAssignedTransactionIds(TransactionId xid, int nsubxids, KnownAssignedXidsRemoveTree(xid, nsubxids, subxids); /* As in ProcArrayEndTransaction, advance latestCompletedXid */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, - max_xid)) - ShmemVariableCache->latestCompletedXid = max_xid; + MaintainLatestCompletedXidRecovery(max_xid); LWLockRelease(ProcArrayLock); } diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 4fdcb07d97b..5b479c8d6e9 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5591,14 +5591,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5610,7 +5611,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 1c063c592ce..ba5d9615c79 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index ceaaa271680..f8411621043 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -412,10 +412,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, Snapshot snapshot = SnapshotAny; /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1437,7 +1437,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index 0cd1160ceb2..e32645d2d3d 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index 96d837485fa..e795f0862fb 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 525d58e7f01..2d821bd817f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -386,6 +386,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -919,6 +920,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --w7fggl772xdywjph Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v9-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PGP.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v11 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 11 +- src/include/access/transam.h | 100 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 39 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 +++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 92 +- src/backend/access/transam/varsup.c | 50 + src/backend/access/transam/xlog.c | 11 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 1021 +++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 ++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 39 files changed, 1626 insertions(+), 608 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index f279edc4734..ef2fcb55a71 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -201,11 +204,15 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +struct GlobalVisState; +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index a91a0c7487d..6ec84b54599 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -54,6 +54,8 @@ #define FullTransactionIdFollowsOrEquals(a, b) ((a).value >= (b).value) #define FullTransactionIdIsValid(x) TransactionIdIsValid(XidFromFullTransactionId(x)) #define InvalidFullTransactionId FullTransactionIdFromEpochAndXid(0, InvalidTransactionId) +#define FirstNormalFullTransactionId FullTransactionIdFromEpochAndXid(0, FirstNormalTransactionId) +#define FullTransactionIdIsNormal(x) FullTransactionIdFollowsOrEquals(x, FirstNormalFullTransactionId) /* * A 64 bit value that contains an epoch and a TransactionId. This is @@ -93,15 +95,48 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ static inline void FullTransactionIdAdvance(FullTransactionId *dest) { dest->value++; + + /* + * In contrast to 32bit XIDs don't step over the "actual" special xids. + * For 64bit xids these can't be reached as part of a wraparound as they + * can in the 32bit case. + */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + /* + * But we do need to step over XIDs that'd appear special only for 32bit + * XIDs. + */ while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) dest->value++; } +/* + * Retreat a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdRetreat(FullTransactionId *dest) +{ + dest->value--; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value--; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -193,8 +228,8 @@ typedef struct VariableCacheData /* * These fields are protected by ProcArrayLock. */ - TransactionId latestCompletedXid; /* newest XID that has committed or - * aborted */ + FullTransactionId latestCompletedFullXid; /* newest full XID that has + * committed or aborted */ /* * These fields are protected by XactTruncationLock @@ -244,6 +279,12 @@ extern void AdvanceOldestClogXid(TransactionId oldest_datfrozenxid); extern bool ForceTransactionIdLimitUpdate(void); extern Oid GetNewObjectId(void); +#ifdef USE_ASSERT_CHECKING +extern void AssertTransactionInAllowableRange(TransactionId xid); +#else +#define AssertTransactionInAllowableRange(xid) ((void)true) +#endif + /* * Some frontend programs include this header. For compilers that emit static * inline functions even when they're unused, that leads to unsatisfied @@ -260,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index b20e2ad4f6a..08f006f782e 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,13 +42,6 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ @@ -56,7 +49,6 @@ struct XidCache #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index a5c7d0c0644..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,41 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_ANALYZE_FLAG 0x04 /* currently running - * analyze */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_ANALYZE_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG -/* Ignore analyze backends */ -#define PROCARRAY_FLAGS_ANALYZE PROCARRAY_FLAGS_DEFAULT | PROCARRAY_ANALYZE_FLAG -/* Ignore both vacuum and analyze backends */ -#define PROCARRAY_FLAGS_VACUUM_ANALYZE PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG | PROCARRAY_ANALYZE_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -88,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 3f0effd5e42..3167305ac00 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index d881f4cd46a..a8804351bee 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1496,6 +1496,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1506,7 +1507,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1595,9 +1597,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 56b35622f1a..659fc4d8697 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1201,7 +1201,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1242,6 +1242,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1261,17 +1272,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index dba10890aab..b25b3e429ed 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index 32ad9e339a2..cf3dba96008 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index 75628e0eb98..9e6376f2c2b 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -990,7 +990,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2209,7 +2209,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index e947addef6b..b59ba02a32f 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -807,6 +807,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -816,13 +822,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index 5d346da84fd..b097e98c3ba 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -928,11 +928,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..4e2178dabab 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -257,31 +257,31 @@ simultaneously, we have one backend take ProcArrayLock and clear the XIDs of multiple processes at once.) ProcArrayEndTransaction also holds the lock while advancing the shared -latestCompletedXid variable. This allows GetSnapshotData to use -latestCompletedXid + 1 as xmax for its snapshot: there can be no +latestCompletedFullXid variable. This allows GetSnapshotData to use +latestCompletedFullXid + 1 as xmax for its snapshot: there can be no transaction >= this xid value that the snapshot needs to consider as completed. In short, then, the rule is that no transaction may exit the set of -currently-running transactions between the time we fetch latestCompletedXid +currently-running transactions between the time we fetch latestCompletedFullXid and the time we finish building our snapshot. However, this restriction only applies to transactions that have an XID --- read-only transactions can end without acquiring ProcArrayLock, since they don't affect anyone -else's snapshot nor latestCompletedXid. +else's snapshot nor latestCompletedFullXid. Transaction start, per se, doesn't have any interlocking with these considerations, since we no longer assign an XID immediately at transaction start. But when we do decide to allocate an XID, GetNewTransactionId must store the new XID into the shared ProcArray before releasing XidGenLock. -This ensures that all top-level XIDs <= latestCompletedXid are either +This ensures that all top-level XIDs <= latestCompletedFullXid are either present in the ProcArray, or not running anymore. (This guarantee doesn't apply to subtransaction XIDs, because of the possibility that there's not room for them in the subxid array; instead we guarantee that they are present or the overflow flag is set.) If a backend released XidGenLock before storing its XID into MyPgXact, then it would be possible for another -backend to allocate and commit a later XID, causing latestCompletedXid to +backend to allocate and commit a later XID, causing latestCompletedFullXid to pass the first backend's XID, before that value became visible in the -ProcArray. That would break GetOldestXmin, as discussed below. +ProcArray. That would break ComputeXidHorizons, as discussed below. We allow GetNewTransactionId to store the XID into MyPgXact->xid (or the subxid array) without taking ProcArrayLock. This was once necessary to @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedFullXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/varsup.c b/src/backend/access/transam/varsup.c index e14b53bf9e3..00b8e4e50d7 100644 --- a/src/backend/access/transam/varsup.c +++ b/src/backend/access/transam/varsup.c @@ -566,3 +566,53 @@ GetNewObjectId(void) return result; } + + +#ifdef USE_ASSERT_CHECKING + +/* + * Assert that xid is between [oldestXid, nextFullXid], which is the range we + * expect XIDs coming from tables etc to be in. + * + * As ShmemVariableCache->oldestXid could change just after this call without + * further precautions, and as a wrapped-around xid could again fall within + * the valid range, this assertion can only detect if something is definitely + * wrong, but not establish correctness. + * + * This intentionally does not expose a return value, to avoid code being + * introduced that depends on the return value. + */ +void +AssertTransactionInAllowableRange(TransactionId xid) +{ + TransactionId oldest_xid; + TransactionId next_xid; + + Assert(TransactionIdIsValid(xid)); + + /* we may see bootstrap / frozen */ + if (!TransactionIdIsNormal(xid)) + return; + + /* + * We can't acquire XidGenLock, as this may be called with XidGenLock + * already held (or with other locks that don't allow XidGenLock to be + * nested). That's ok for our purposes though, since we already rely on + * 32bit reads to be atomic. While nextFullXid is 64 bit, we only look at + * the lower 32bit, so a skewed read doesn't hurt. + * + * There's no increased danger of falling outside [oldest, next] by + * accessing them without a lock. xid needs to have been created with + * GetNewTransactionId() in the originating session, and the locks there + * pair with the memory barrier below. We do however accept xid to be <= + * to next_xid, instead of just <, as xid could be from the procarray, + * before we see the updated nextFullXid value. + */ + pg_memory_barrier(); + oldest_xid = ShmemVariableCache->oldestXid; + next_xid = XidFromFullTransactionId(ShmemVariableCache->nextFullXid); + + Assert(TransactionIdFollowsOrEquals(xid, oldest_xid) || + TransactionIdPrecedesOrEquals(xid, next_xid)); +} +#endif diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 0a97b1d37fb..4ec414e93ef 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -7866,10 +7866,11 @@ StartupXLOG(void) XLogCtl->lastSegSwitchTime = (pg_time_t) time(NULL); XLogCtl->lastSegSwitchLSN = EndOfLog; - /* also initialize latestCompletedXid, to nextXid - 1 */ + /* also initialize latestCompletedFullXid, to nextFullXid - 1 */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); - ShmemVariableCache->latestCompletedXid = XidFromFullTransactionId(ShmemVariableCache->nextFullXid); - TransactionIdRetreat(ShmemVariableCache->latestCompletedXid); + ShmemVariableCache->latestCompletedFullXid = + ShmemVariableCache->nextFullXid; + FullTransactionIdRetreat(&ShmemVariableCache->latestCompletedFullXid); LWLockRelease(ProcArrayLock); /* @@ -9099,7 +9100,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9459,7 +9460,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index 924ef37c816..34b71b6c1c5 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1056,7 +1056,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index aec885e9871..158b2f3d73b 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index 5e2210dd7bd..fd370d52b66 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2116,9 +2116,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2131,10 +2132,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index b4485335644..c011387ba90 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* familiy of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedFullXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -175,6 +327,10 @@ static void KnownAssignedXidsReset(void); static inline void ProcArrayEndTransactionInternal(PGPROC *proc, PGXACT *pgxact, TransactionId latestXid); static void ProcArrayGroupClearXid(PGPROC *proc, TransactionId latestXid); +static void MaintainLatestCompletedXid(TransactionId latestXid); +static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); + +static inline FullTransactionId FullXidViaRelative(FullTransactionId rel, TransactionId xid); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -349,9 +505,7 @@ ProcArrayRemove(PGPROC *proc, TransactionId latestXid) Assert(TransactionIdIsValid(allPgXact[proc->pgprocno].xid)); /* Advance global latestCompletedXid while holding the lock */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, - latestXid)) - ShmemVariableCache->latestCompletedXid = latestXid; + MaintainLatestCompletedXid(latestXid); } else { @@ -464,9 +618,7 @@ ProcArrayEndTransactionInternal(PGPROC *proc, PGXACT *pgxact, pgxact->overflowed = false; /* Also advance global latestCompletedXid while holding the lock */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, - latestXid)) - ShmemVariableCache->latestCompletedXid = latestXid; + MaintainLatestCompletedXid(latestXid); } /* @@ -621,6 +773,58 @@ ProcArrayClearTransaction(PGPROC *proc) pgxact->overflowed = false; } +/* + * Update ShmemVariableCache->latestCompletedFullXid to point to latestXid if + * currently older. + */ +static void +MaintainLatestCompletedXid(TransactionId latestXid) +{ + FullTransactionId cur_latest = ShmemVariableCache->latestCompletedFullXid; + + Assert(LWLockHeldByMe(ProcArrayLock)); + Assert(FullTransactionIdIsValid(cur_latest)); + + if (TransactionIdPrecedes(XidFromFullTransactionId(cur_latest), latestXid)) + { + ShmemVariableCache->latestCompletedFullXid = + FullXidViaRelative(cur_latest, latestXid); + } + + Assert(IsBootstrapProcessingMode() || + FullTransactionIdIsNormal(ShmemVariableCache->latestCompletedFullXid)); +} + +/* + * Same as MaintainLatestCompletedXid, except for use during WAL replay. + */ +static void +MaintainLatestCompletedXidRecovery(TransactionId latestXid) +{ + FullTransactionId cur_latest = ShmemVariableCache->latestCompletedFullXid; + FullTransactionId rel; + + Assert(AmStartupProcess() || !IsUnderPostmaster); + Assert(LWLockHeldByMe(ProcArrayLock)); + + /* + * Need a FullTransactionId to compare latestXid with. Can't rely on + * latestCompletedFullXid to be initialized in recovery. But in recovery + * it's safe to access nextFullXid without a lock for the startup process. + */ + rel = ShmemVariableCache->nextFullXid; + Assert(FullTransactionIdIsValid(ShmemVariableCache->nextFullXid)); + + if (!FullTransactionIdIsValid(cur_latest) || + TransactionIdPrecedes(XidFromFullTransactionId(cur_latest), latestXid)) + { + ShmemVariableCache->latestCompletedFullXid = + FullXidViaRelative(rel, latestXid); + } + + Assert(FullTransactionIdIsNormal(ShmemVariableCache->latestCompletedFullXid)); +} + /* * ProcArrayInitRecovery -- initialize recovery xid mgmt environment * @@ -841,7 +1045,7 @@ ProcArrayApplyRecoveryInfo(RunningTransactions running) * Now we've got the running xids we need to set the global values that * are used to track snapshots as they evolve further. * - * - latestCompletedXid which will be the xmax for snapshots + * - latestCompletedFullXid which will be the xmax for snapshots * - lastOverflowedXid which shows whether snapshots overflow * - nextXid * @@ -867,14 +1071,11 @@ ProcArrayApplyRecoveryInfo(RunningTransactions running) /* * If a transaction wrote a commit record in the gap between taking and - * logging the snapshot then latestCompletedXid may already be higher than - * the value from the snapshot, so check before we use the incoming value. + * logging the snapshot then latestCompletedFullXid may already be higher + * than the value from the snapshot, so check before we use the incoming + * value. It also might not yet be set at all. */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, - running->latestCompletedXid)) - ShmemVariableCache->latestCompletedXid = running->latestCompletedXid; - - Assert(TransactionIdIsNormal(ShmemVariableCache->latestCompletedXid)); + MaintainLatestCompletedXidRecovery(running->latestCompletedXid); LWLockRelease(ProcArrayLock); @@ -1048,10 +1249,11 @@ TransactionIdIsInProgress(TransactionId xid) LWLockAcquire(ProcArrayLock, LW_SHARED); /* - * Now that we have the lock, we can check latestCompletedXid; if the + * Now that we have the lock, we can check latestCompletedFullXid; if the * target Xid is after that, it's surely still running. */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, xid)) + if (TransactionIdPrecedes(XidFromFullTransactionId(ShmemVariableCache->latestCompletedFullXid), + xid)) { LWLockRelease(ProcArrayLock); xc_by_latest_xid_inc(); @@ -1248,159 +1450,183 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See ComputedXidHorizonsResult for the various computed horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedFullXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); - /* - * We initialize the MIN() calculation with latestCompletedXid + 1. This - * is a lower bound for the XIDs that might appear in the ProcArray later, - * and so protects us against overestimating the result due to future - * additions. - */ - result = ShmemVariableCache->latestCompletedXid; - Assert(TransactionIdIsNormal(result)); - TransactionIdAdvance(result); + h->latest_completed = ShmemVariableCache->latestCompletedFullXid; - for (index = 0; index < arrayP->numProcs; index++) + /* + * We initialize the MIN() calculation with latestCompletedFullXid + 1. + * This is a lower bound for the XIDs that might appear in the ProcArray + * later, and so protects us against overestimating the result due to + * future additions. + */ + { + TransactionId initial; + + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1412,34 +1638,143 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1490,12 +1825,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1506,11 +1839,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; + FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1554,13 +1888,16 @@ GetSnapshotData(Snapshot snapshot) */ LWLockAcquire(ProcArrayLock, LW_SHARED); + latest_completed = ShmemVariableCache->latestCompletedFullXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ - xmax = ShmemVariableCache->latestCompletedXid; - Assert(TransactionIdIsNormal(xmax)); + xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); + Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1589,12 +1926,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1710,34 +2041,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidViaRelative(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidViaRelative(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidViaRelative(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -1984,7 +2359,7 @@ GetRunningTransactionData(void) LWLockAcquire(ProcArrayLock, LW_SHARED); LWLockAcquire(XidGenLock, LW_SHARED); - latestCompletedXid = ShmemVariableCache->latestCompletedXid; + latestCompletedXid = XidFromFullTransactionId(ShmemVariableCache->latestCompletedFullXid); oldestRunningXid = XidFromFullTransactionId(ShmemVariableCache->nextFullXid); @@ -3207,9 +3582,7 @@ XidCacheRemoveRunningXids(TransactionId xid, elog(WARNING, "did not find subXID %u in MyProc", xid); /* Also advance global latestCompletedXid while holding the lock */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, - latestXid)) - ShmemVariableCache->latestCompletedXid = latestXid; + MaintainLatestCompletedXid(latestXid); LWLockRelease(ProcArrayLock); } @@ -3236,6 +3609,276 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + GlobalVisSharedRels.maybe_needed = + FullXidViaRelative(horizons.latest_completed, + horizons.shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidViaRelative(horizons.latest_completed, + horizons.catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidViaRelative(horizons.latest_completed, + horizons.data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidViaRelative(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + +/* + * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it + * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). + * + * Be very careful about when to use this function. It can only safely be used + * when there is a guarantee that xid is within MaxTransactionId / 2 xids of + * rel. That e.g. can be guaranteed if the the caller assures a snapshot is + * held by the backend and xid is from a table (where vacuum/freezing ensures + * the xid has to be within that range), or if xid is from the procarray and + * prevents xid wraparound that way. + */ +static inline FullTransactionId +FullXidViaRelative(FullTransactionId rel, TransactionId xid) +{ + TransactionId rel_xid = XidFromFullTransactionId(rel); + + Assert(TransactionIdIsValid(xid)); + Assert(TransactionIdIsValid(rel_xid)); + + /* not guaranteed to find issues, but likely to catch mistakes */ + AssertTransactionInAllowableRange(xid); + + return FullTransactionIdFromU64(U64FromFullTransactionId(rel) + + (int32) (xid - rel_xid)); +} + /* ---------------------------------------------- * KnownAssignedTransactionIds sub-module @@ -3388,9 +4031,7 @@ ExpireTreeKnownAssignedTransactionIds(TransactionId xid, int nsubxids, KnownAssignedXidsRemoveTree(xid, nsubxids, subxids); /* As in ProcArrayEndTransaction, advance latestCompletedXid */ - if (TransactionIdPrecedes(ShmemVariableCache->latestCompletedXid, - max_xid)) - ShmemVariableCache->latestCompletedXid = max_xid; + MaintainLatestCompletedXidRecovery(max_xid); LWLockRelease(ProcArrayLock); } diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index be08eb48148..2d4ec92c2a1 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5783,14 +5783,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5802,7 +5803,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index e4d501a85d1..76306976c2a 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -419,10 +419,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1441,7 +1441,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index 68d580ed1e0..37206c50a21 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --kdyesixubxkl3b7r Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v11-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v12 2/7] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 39 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 573 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index f279edc4734..232db64ecdf 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -201,11 +204,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 9502217932e..cf221582cd0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index b20e2ad4f6a..08f006f782e 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,13 +42,6 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ @@ -56,7 +49,6 @@ struct XidCache #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index a5c7d0c0644..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,41 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_ANALYZE_FLAG 0x04 /* currently running - * analyze */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_ANALYZE_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG -/* Ignore analyze backends */ -#define PROCARRAY_FLAGS_ANALYZE PROCARRAY_FLAGS_DEFAULT | PROCARRAY_ANALYZE_FLAG -/* Ignore both vacuum and analyze backends */ -#define PROCARRAY_FLAGS_VACUUM_ANALYZE PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG | PROCARRAY_ANALYZE_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -88,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 3f0effd5e42..3167305ac00 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index d881f4cd46a..a8804351bee 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1496,6 +1496,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1506,7 +1507,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1595,9 +1597,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 8f2e5379210..8c0d601e6d9 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1202,7 +1202,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1243,6 +1243,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1262,17 +1273,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index dba10890aab..b25b3e429ed 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index 32ad9e339a2..cf3dba96008 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index 70bac0052fc..d18b2722693 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2316,7 +2316,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index d65f4357cc8..922774bcf3b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -807,6 +807,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -816,13 +822,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1275,14 +1280,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index 5d346da84fd..b097e98c3ba 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -928,11 +928,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index c06e52f9cd0..4e2178dabab 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedFullXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedFullXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index e1a043763cf..4420d59f26d 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9100,7 +9100,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9460,7 +9460,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index 924ef37c816..34b71b6c1c5 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1056,7 +1056,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index 5e2210dd7bd..fd370d52b66 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2116,9 +2116,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2131,10 +2132,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 82798760752..20115e2f63f 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedFullXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -179,6 +331,7 @@ static void MaintainLatestCompletedXid(TransactionId latestXid); static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidViaRelative(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1299,159 +1452,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedFullXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedFullXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedFullXid; + /* * We initialize the MIN() calculation with latestCompletedFullXid + 1. * This is a lower bound for the XIDs that might appear in the ProcArray * later, and so protects us against overestimating the result due to * future additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedFullXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1463,34 +1648,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1541,12 +1838,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1557,12 +1852,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1607,13 +1902,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedFullXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedFullXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1642,12 +1939,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1763,34 +2054,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidViaRelative(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidViaRelative(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidViaRelative(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3288,6 +3623,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidViaRelative(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidViaRelative(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidViaRelative(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidViaRelative(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index e4d501a85d1..76306976c2a 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -419,10 +419,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1441,7 +1441,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index 68d580ed1e0..37206c50a21 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --pyx7ptyhkbjgwa43 Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v12-0003-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. @ 2020-07-15 22:35 Andres Freund <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: Andres Freund @ 2020-07-15 22:35 UTC (permalink / raw) To make GetSnapshotData() more scalable, it cannot not look at at each proc's xmin (see Discussion link below). Due to the frequency at which xmins are updated, that just does not scale. Without accessing xmins GetSnapshotData() cannot calculate accurate thresholds as it has so far. But we don't really have to: The horizons don't actually change that much between GetSnapshotData() calls. Nor are the horizons actually used every time a snapshot is called. The use of RecentGlobal[Data]Xmin to decide whether a row version could be removed has been replaces with new GlobalVisTest* functions. These use two thresholds to determine whether a row can be pruned: 1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed are definitely still visible. 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can definitely be removed GetSnapshotData() updates definitely_needed to be the xmin of the computed snapshot. When testing whether a row can be removed (with GlobalVisTestIsRemovableXid()) and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID < definitely_needed) the boundaries can be recomputed to be more accurate. As it is not cheap to compute accurate boundaries, we limit the number of times that happens in short succession. As the boundaries used by GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated byGetSnapshotData()), it is likely that further test can benefit from an earlier computation of accurate horizons. To avoid regressing performance when old_snapshot_threshold is set (as that requires an accurate horizon to be computed), heap_page_prune_opt() doesn't unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the computation of the limited horizon, and the triggering of errors (with SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove tuples. Subsequent commits will take further advantage of the fact that GetSnapshotData() will not need to access xmins anymore. Note: This contains a workaround in heap_page_prune_opt() to keep the snapshot_too_old tests working. While that workaround is ugly, the tests currently are not meaningful, and it seems best to address them separately. Author: Andres Freund Reviewed-By: Robert Haas, Thomas Munro, David Rowley Discussion: https://postgr.es/m/[email protected] --- src/include/access/ginblock.h | 4 +- src/include/access/heapam.h | 10 +- src/include/access/transam.h | 79 +- src/include/storage/bufpage.h | 6 - src/include/storage/proc.h | 8 - src/include/storage/procarray.h | 32 +- src/include/utils/snapmgr.h | 37 +- src/include/utils/snapshot.h | 6 + src/backend/access/gin/ginvacuum.c | 26 + src/backend/access/gist/gistutil.c | 8 +- src/backend/access/gist/gistxlog.c | 10 +- src/backend/access/heap/heapam.c | 15 +- src/backend/access/heap/heapam_handler.c | 24 +- src/backend/access/heap/heapam_visibility.c | 79 +- src/backend/access/heap/pruneheap.c | 207 ++++- src/backend/access/heap/vacuumlazy.c | 24 +- src/backend/access/index/indexam.c | 3 +- src/backend/access/nbtree/README | 10 +- src/backend/access/nbtree/nbtpage.c | 4 +- src/backend/access/nbtree/nbtree.c | 28 +- src/backend/access/nbtree/nbtxlog.c | 10 +- src/backend/access/spgist/spgvacuum.c | 6 +- src/backend/access/transam/README | 78 +- src/backend/access/transam/xlog.c | 4 +- src/backend/commands/analyze.c | 2 +- src/backend/commands/vacuum.c | 41 +- src/backend/postmaster/autovacuum.c | 4 + src/backend/replication/logical/launcher.c | 4 + src/backend/replication/walreceiver.c | 17 +- src/backend/replication/walsender.c | 15 +- src/backend/storage/ipc/procarray.c | 902 ++++++++++++++++---- src/backend/utils/adt/selfuncs.c | 20 +- src/backend/utils/init/postinit.c | 4 + src/backend/utils/time/snapmgr.c | 258 +++--- contrib/amcheck/verify_nbtree.c | 8 +- contrib/pg_visibility/pg_visibility.c | 18 +- contrib/pgstattuple/pgstatapprox.c | 2 +- src/tools/pgindent/typedefs.list | 2 + 38 files changed, 1449 insertions(+), 566 deletions(-) diff --git a/src/include/access/ginblock.h b/src/include/access/ginblock.h index 3f64fd572e3..fe66a95226b 100644 --- a/src/include/access/ginblock.h +++ b/src/include/access/ginblock.h @@ -12,6 +12,7 @@ #include "access/transam.h" #include "storage/block.h" +#include "storage/bufpage.h" #include "storage/itemptr.h" #include "storage/off.h" @@ -134,8 +135,7 @@ typedef struct GinMetaPageData */ #define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid ) #define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid) -#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \ - && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin))) +extern bool GinPageIsRecyclable(Page page); /* * We use our own ItemPointerGet(BlockNumber|OffsetNumber) diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h index b31de389106..ebb79428d16 100644 --- a/src/include/access/heapam.h +++ b/src/include/access/heapam.h @@ -172,9 +172,12 @@ extern TransactionId heap_compute_xid_horizon_for_tuples(Relation rel, int nitems); /* in heap/pruneheap.c */ +struct GlobalVisState; extern void heap_page_prune_opt(Relation relation, Buffer buffer); extern int heap_page_prune(Relation relation, Buffer buffer, - TransactionId OldestXmin, + struct GlobalVisState *vistest, + TransactionId limited_oldest_xmin, + TimestampTz limited_oldest_ts, bool report_stats, TransactionId *latestRemovedXid); extern void heap_page_prune_execute(Buffer buffer, OffsetNumber *redirected, int nredirected, @@ -195,11 +198,14 @@ extern TM_Result HeapTupleSatisfiesUpdate(HeapTuple stup, CommandId curcid, Buffer buffer); extern HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple stup, TransactionId OldestXmin, Buffer buffer); +extern HTSV_Result HeapTupleSatisfiesVacuumHorizon(HeapTuple stup, Buffer buffer, + TransactionId *dead_after); extern void HeapTupleSetHintBits(HeapTupleHeader tuple, Buffer buffer, uint16 infomask, TransactionId xid); extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple); extern bool XidInMVCCSnapshot(TransactionId xid, Snapshot snapshot); -extern bool HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin); +extern bool HeapTupleIsSurelyDead(struct GlobalVisState *vistest, + HeapTuple htup); /* * To avoid leaking too much knowledge about reorderbuffer implementation diff --git a/src/include/access/transam.h b/src/include/access/transam.h index 8db326ad1b5..b32044153b0 100644 --- a/src/include/access/transam.h +++ b/src/include/access/transam.h @@ -95,15 +95,6 @@ FullTransactionIdFromU64(uint64 value) (dest) = FirstNormalTransactionId; \ } while(0) -/* advance a FullTransactionId variable, stepping over special XIDs */ -static inline void -FullTransactionIdAdvance(FullTransactionId *dest) -{ - dest->value++; - while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) - dest->value++; -} - /* * Retreat a FullTransactionId variable, stepping over xids that would appear * to be special only when viewed as 32bit XIDs. @@ -129,6 +120,23 @@ FullTransactionIdRetreat(FullTransactionId *dest) dest->value--; } +/* + * Advance a FullTransactionId variable, stepping over xids that would appear + * to be special only when viewed as 32bit XIDs. + */ +static inline void +FullTransactionIdAdvance(FullTransactionId *dest) +{ + dest->value++; + + /* see FullTransactionIdAdvance() */ + if (FullTransactionIdPrecedes(*dest, FirstNormalFullTransactionId)) + return; + + while (XidFromFullTransactionId(*dest) < FirstNormalTransactionId) + dest->value++; +} + /* back up a transaction ID variable, handling wraparound correctly */ #define TransactionIdRetreat(dest) \ do { \ @@ -293,6 +301,59 @@ ReadNewTransactionId(void) return XidFromFullTransactionId(ReadNextFullTransactionId()); } +/* return transaction ID backed up by amount, handling wraparound correctly */ +static inline TransactionId +TransactionIdRetreatedBy(TransactionId xid, uint32 amount) +{ + xid -= amount; + + while (xid < FirstNormalTransactionId) + xid--; + + return xid; +} + +/* return the older of the two IDs */ +static inline TransactionId +TransactionIdOlder(TransactionId a, TransactionId b) +{ + if (!TransactionIdIsValid(a)) + return b; + + if (!TransactionIdIsValid(b)) + return a; + + if (TransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the older of the two IDs, assuming they're both normal */ +static inline TransactionId +NormalTransactionIdOlder(TransactionId a, TransactionId b) +{ + Assert(TransactionIdIsNormal(a)); + Assert(TransactionIdIsNormal(b)); + if (NormalTransactionIdPrecedes(a, b)) + return a; + return b; +} + +/* return the newer of the two IDs */ +static inline FullTransactionId +FullTransactionIdNewer(FullTransactionId a, FullTransactionId b) +{ + if (!FullTransactionIdIsValid(a)) + return b; + + if (!FullTransactionIdIsValid(b)) + return a; + + if (FullTransactionIdFollows(a, b)) + return a; + return b; +} + #endif /* FRONTEND */ #endif /* TRANSAM_H */ diff --git a/src/include/storage/bufpage.h b/src/include/storage/bufpage.h index 3f88683a059..51b8f994ac0 100644 --- a/src/include/storage/bufpage.h +++ b/src/include/storage/bufpage.h @@ -389,12 +389,6 @@ PageValidateSpecialPointer(Page page) #define PageClearAllVisible(page) \ (((PageHeader) (page))->pd_flags &= ~PD_ALL_VISIBLE) -#define PageIsPrunable(page, oldestxmin) \ -( \ - AssertMacro(TransactionIdIsNormal(oldestxmin)), \ - TransactionIdIsValid(((PageHeader) (page))->pd_prune_xid) && \ - TransactionIdPrecedes(((PageHeader) (page))->pd_prune_xid, oldestxmin) \ -) #define PageSetPrunable(page, xid) \ do { \ Assert(TransactionIdIsNormal(xid)); \ diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h index 5ceb2494bae..52ff43cabaa 100644 --- a/src/include/storage/proc.h +++ b/src/include/storage/proc.h @@ -42,20 +42,12 @@ struct XidCache /* * Flags for PGXACT->vacuumFlags - * - * Note: If you modify these flags, you need to modify PROCARRAY_XXX flags - * in src/include/storage/procarray.h. - * - * PROC_RESERVED may later be assigned for use in vacuumFlags, but its value is - * used for PROCARRAY_SLOTS_XMIN in procarray.h, so GetOldestXmin won't be able - * to match and ignore processes with this flag set. */ #define PROC_IS_AUTOVACUUM 0x01 /* is it an autovac worker? */ #define PROC_IN_VACUUM 0x02 /* currently running lazy vacuum */ #define PROC_VACUUM_FOR_WRAPAROUND 0x08 /* set by autovac only */ #define PROC_IN_LOGICAL_DECODING 0x10 /* currently doing logical * decoding outside xact */ -#define PROC_RESERVED 0x20 /* reserved for procarray */ /* flags reset at EOXact */ #define PROC_VACUUM_STATE_MASK \ diff --git a/src/include/storage/procarray.h b/src/include/storage/procarray.h index 01040d76e12..ea8a876ca45 100644 --- a/src/include/storage/procarray.h +++ b/src/include/storage/procarray.h @@ -20,34 +20,6 @@ #include "utils/snapshot.h" -/* - * These are to implement PROCARRAY_FLAGS_XXX - * - * Note: These flags are cloned from PROC_XXX flags in src/include/storage/proc.h - * to avoid forcing to include proc.h when including procarray.h. So if you modify - * PROC_XXX flags, you need to modify these flags. - */ -#define PROCARRAY_VACUUM_FLAG 0x02 /* currently running lazy - * vacuum */ -#define PROCARRAY_LOGICAL_DECODING_FLAG 0x10 /* currently doing logical - * decoding outside xact */ - -#define PROCARRAY_SLOTS_XMIN 0x20 /* replication slot xmin, - * catalog_xmin */ -/* - * Only flags in PROCARRAY_PROC_FLAGS_MASK are considered when matching - * PGXACT->vacuumFlags. Other flags are used for different purposes and - * have no corresponding PROC flag equivalent. - */ -#define PROCARRAY_PROC_FLAGS_MASK (PROCARRAY_VACUUM_FLAG | \ - PROCARRAY_LOGICAL_DECODING_FLAG) - -/* Use the following flags as an input "flags" to GetOldestXmin function */ -/* Consider all backends except for logical decoding ones which manage xmin separately */ -#define PROCARRAY_FLAGS_DEFAULT PROCARRAY_LOGICAL_DECODING_FLAG -/* Ignore vacuum backends */ -#define PROCARRAY_FLAGS_VACUUM PROCARRAY_FLAGS_DEFAULT | PROCARRAY_VACUUM_FLAG - extern Size ProcArrayShmemSize(void); extern void CreateSharedProcArray(void); extern void ProcArrayAdd(PGPROC *proc); @@ -81,9 +53,11 @@ extern RunningTransactions GetRunningTransactionData(void); extern bool TransactionIdIsInProgress(TransactionId xid); extern bool TransactionIdIsActive(TransactionId xid); -extern TransactionId GetOldestXmin(Relation rel, int flags); +extern TransactionId GetOldestNonRemovableTransactionId(Relation rel); +extern TransactionId GetOldestTransactionIdConsideredRunning(void); extern TransactionId GetOldestActiveTransactionId(void); extern TransactionId GetOldestSafeDecodingTransactionId(bool catalogOnly); +extern void GetReplicationHorizons(TransactionId *slot_xmin, TransactionId *catalog_xmin); extern VirtualTransactionId *GetVirtualXIDsDelayingChkpt(int *nvxids); extern bool HaveVirtualXIDsDelayingChkpt(VirtualTransactionId *vxids, int nvxids); diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h index ffb4ba3adfb..b6b403e2931 100644 --- a/src/include/utils/snapmgr.h +++ b/src/include/utils/snapmgr.h @@ -52,13 +52,12 @@ extern Size SnapMgrShmemSize(void); extern void SnapMgrInit(void); extern TimestampTz GetSnapshotCurrentTimestamp(void); extern TimestampTz GetOldSnapshotThresholdTimestamp(void); +extern void SnapshotTooOldMagicForTest(void); extern bool FirstSnapshotSet; extern PGDLLIMPORT TransactionId TransactionXmin; extern PGDLLIMPORT TransactionId RecentXmin; -extern PGDLLIMPORT TransactionId RecentGlobalXmin; -extern PGDLLIMPORT TransactionId RecentGlobalDataXmin; /* Variables representing various special snapshot semantics */ extern PGDLLIMPORT SnapshotData SnapshotSelfData; @@ -78,11 +77,12 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; /* * Similarly, some initialization is required for a NonVacuumable snapshot. - * The caller must supply the xmin horizon to use (e.g., RecentGlobalXmin). + * The caller must supply the visibility cutoff state to use (c.f. + * GlobalVisTestFor()). */ -#define InitNonVacuumableSnapshot(snapshotdata, xmin_horizon) \ +#define InitNonVacuumableSnapshot(snapshotdata, vistestp) \ ((snapshotdata).snapshot_type = SNAPSHOT_NON_VACUUMABLE, \ - (snapshotdata).xmin = (xmin_horizon)) + (snapshotdata).vistest = (vistestp)) /* * Similarly, some initialization is required for SnapshotToast. We need @@ -98,6 +98,11 @@ extern PGDLLIMPORT SnapshotData CatalogSnapshotData; ((snapshot)->snapshot_type == SNAPSHOT_MVCC || \ (snapshot)->snapshot_type == SNAPSHOT_HISTORIC_MVCC) +static inline bool +OldSnapshotThresholdActive(void) +{ + return old_snapshot_threshold >= 0; +} extern Snapshot GetTransactionSnapshot(void); extern Snapshot GetLatestSnapshot(void); @@ -121,8 +126,6 @@ extern void UnregisterSnapshot(Snapshot snapshot); extern Snapshot RegisterSnapshotOnOwner(Snapshot snapshot, ResourceOwner owner); extern void UnregisterSnapshotFromOwner(Snapshot snapshot, ResourceOwner owner); -extern FullTransactionId GetFullRecentGlobalXmin(void); - extern void AtSubCommit_Snapshot(int level); extern void AtSubAbort_Snapshot(int level); extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin); @@ -131,13 +134,29 @@ extern void ImportSnapshot(const char *idstr); extern bool XactHasExportedSnapshots(void); extern void DeleteAllExportedSnapshotFiles(void); extern bool ThereAreNoPriorRegisteredSnapshots(void); -extern TransactionId TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation); +extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts); +extern void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit); extern void MaintainOldSnapshotTimeMapping(TimestampTz whenTaken, TransactionId xmin); extern char *ExportSnapshot(Snapshot snapshot); +/* + * These live in procarray.c because they're intimately linked to the + * procarray contents, but thematically they better fit into snapmgr.h. + */ +typedef struct GlobalVisState GlobalVisState; +extern GlobalVisState *GlobalVisTestFor(Relation rel); +extern bool GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid); +extern bool GlobalVisTestIsRemovableFullXid(GlobalVisState *state, FullTransactionId fxid); +extern FullTransactionId GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state); +extern TransactionId GlobalVisTestNonRemovableHorizon(GlobalVisState *state); +extern bool GlobalVisCheckRemovableXid(Relation rel, TransactionId xid); +extern bool GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid); + /* * Utility functions for implementing visibility routines in table AMs. */ diff --git a/src/include/utils/snapshot.h b/src/include/utils/snapshot.h index 4796edb63aa..35b1f05bea6 100644 --- a/src/include/utils/snapshot.h +++ b/src/include/utils/snapshot.h @@ -192,6 +192,12 @@ typedef struct SnapshotData */ uint32 speculativeToken; + /* + * For SNAPSHOT_NON_VACUUMABLE (and hopefully more in the future) this is + * used to determine whether row could be vacuumed. + */ + struct GlobalVisState *vistest; + /* * Book-keeping information, used by the snapshot manager */ diff --git a/src/backend/access/gin/ginvacuum.c b/src/backend/access/gin/ginvacuum.c index 8ae4fd95a7b..9cd6638df62 100644 --- a/src/backend/access/gin/ginvacuum.c +++ b/src/backend/access/gin/ginvacuum.c @@ -793,3 +793,29 @@ ginvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) return stats; } + +/* + * Return whether Page can safely be recycled. + */ +bool +GinPageIsRecyclable(Page page) +{ + TransactionId delete_xid; + + if (PageIsNew(page)) + return true; + + if (!GinPageIsDeleted(page)) + return false; + + delete_xid = GinPageGetDeleteXid(page); + + if (!TransactionIdIsValid(delete_xid)) + return true; + + /* + * If no backend still could view delete_xid as in running, all scans + * concurrent with ginDeletePage() must have finished. + */ + return GlobalVisCheckRemovableXid(NULL, delete_xid); +} diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index 765329bbcd4..bfda7fbe3d5 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -891,15 +891,13 @@ gistPageRecyclable(Page page) * As long as that can happen, we must keep the deleted page around as * a tombstone. * - * Compare the deletion XID with RecentGlobalXmin. If deleteXid < - * RecentGlobalXmin, then no scan that's still in progress could have + * For that check if the deletion XID could still be visible to + * anyone. If not, then no scan that's still in progress could have * seen its downlink, and we can recycle it. */ FullTransactionId deletexid_full = GistPageGetDeleteXid(page); - FullTransactionId recentxmin_full = GetFullRecentGlobalXmin(); - if (FullTransactionIdPrecedes(deletexid_full, recentxmin_full)) - return true; + return GlobalVisIsRemovableFullXid(NULL, deletexid_full); } return false; } diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c index 7b5d1e98b70..a63b05388c5 100644 --- a/src/backend/access/gist/gistxlog.c +++ b/src/backend/access/gist/gistxlog.c @@ -387,11 +387,11 @@ gistRedoPageReuse(XLogReaderState *record) * PAGE_REUSE records exist to provide a conflict point when we reuse * pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's deleteXid. The deleteXid < - * RecentGlobalXmin test in gistPageRecyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's deleteXid. The + * GlobalVisIsRemovableFullXid(deleteXid) test in gistPageRecyclable() + * conceptually mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 00169006fb1..0a89e741a15 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -1517,6 +1517,7 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, bool at_chain_start; bool valid; bool skip; + GlobalVisState *vistest = NULL; /* If this is not the first call, previous call returned a (live!) tuple */ if (all_dead) @@ -1527,7 +1528,8 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, at_chain_start = first_call; skip = !first_call; - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); Assert(BufferGetBlockNumber(buffer) == blkno); /* Scan through possible multiple members of HOT-chain */ @@ -1616,9 +1618,14 @@ heap_hot_search_buffer(ItemPointer tid, Relation relation, Buffer buffer, * Note: if you change the criterion here for what is "dead", fix the * planner's get_actual_variable_range() function to match. */ - if (all_dead && *all_dead && - !HeapTupleIsSurelyDead(heapTuple, RecentGlobalXmin)) - *all_dead = false; + if (all_dead && *all_dead) + { + if (!vistest) + vistest = GlobalVisTestFor(relation); + + if (!HeapTupleIsSurelyDead(vistest, heapTuple)) + *all_dead = false; + } /* * Check to see if HOT chain continues past this tuple; if so fetch diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index 267a6ee25a7..e3e41fb7516 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -1203,7 +1203,7 @@ heapam_index_build_range_scan(Relation heapRelation, /* okay to ignore lazy VACUUMs here */ if (!IsBootstrapProcessingMode() && !indexInfo->ii_Concurrent) - OldestXmin = GetOldestXmin(heapRelation, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(heapRelation); if (!scan) { @@ -1244,6 +1244,17 @@ heapam_index_build_range_scan(Relation heapRelation, hscan = (HeapScanDesc) scan; + /* + * Must have called GetOldestNonRemovableTransactionId() if using + * SnapshotAny. Shouldn't have for an MVCC snapshot. (It's especially + * worth checking this for parallel builds, since ambuild routines that + * support parallel builds must work these details out for themselves.) + */ + Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); + Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : + !TransactionIdIsValid(OldestXmin)); + Assert(snapshot == SnapshotAny || !anyvisible); + /* Publish number of blocks to scan */ if (progress) { @@ -1263,17 +1274,6 @@ heapam_index_build_range_scan(Relation heapRelation, nblocks); } - /* - * Must call GetOldestXmin() with SnapshotAny. Should never call - * GetOldestXmin() with MVCC snapshot. (It's especially worth checking - * this for parallel builds, since ambuild routines that support parallel - * builds must work these details out for themselves.) - */ - Assert(snapshot == SnapshotAny || IsMVCCSnapshot(snapshot)); - Assert(snapshot == SnapshotAny ? TransactionIdIsValid(OldestXmin) : - !TransactionIdIsValid(OldestXmin)); - Assert(snapshot == SnapshotAny || !anyvisible); - /* set our scan endpoints */ if (!allow_sync) heap_setscanlimits(scan, start_blockno, numblocks); diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c index c77128087cf..f117ee160a3 100644 --- a/src/backend/access/heap/heapam_visibility.c +++ b/src/backend/access/heap/heapam_visibility.c @@ -1154,19 +1154,56 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot, * we mainly want to know is if a tuple is potentially visible to *any* * running transaction. If so, it can't be removed yet by VACUUM. * - * OldestXmin is a cutoff XID (obtained from GetOldestXmin()). Tuples - * deleted by XIDs >= OldestXmin are deemed "recently dead"; they might - * still be visible to some open transaction, so we can't remove them, - * even if we see that the deleting transaction has committed. + * OldestXmin is a cutoff XID (obtained from + * GetOldestNonRemovableTransactionId()). Tuples deleted by XIDs >= + * OldestXmin are deemed "recently dead"; they might still be visible to some + * open transaction, so we can't remove them, even if we see that the deleting + * transaction has committed. */ HTSV_Result HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, Buffer buffer) +{ + TransactionId dead_after = InvalidTransactionId; + HTSV_Result res; + + res = HeapTupleSatisfiesVacuumHorizon(htup, buffer, &dead_after); + + if (res == HEAPTUPLE_RECENTLY_DEAD) + { + Assert(TransactionIdIsValid(dead_after)); + + if (TransactionIdPrecedes(dead_after, OldestXmin)) + res = HEAPTUPLE_DEAD; + } + else + Assert(!TransactionIdIsValid(dead_after)); + + return res; +} + +/* + * Work horse for HeapTupleSatisfiesVacuum and similar routines. + * + * In contrast to HeapTupleSatisfiesVacuum this routine, when encountering a + * tuple that could still be visible to some backend, stores the xid that + * needs to be compared with the horizon in *dead_after, and returns + * HEAPTUPLE_RECENTLY_DEAD. The caller then can perform the comparison with + * the horizon. This is e.g. useful when comparing with different horizons. + * + * Note: HEAPTUPLE_DEAD can still be returned here, e.g. if the inserting + * transaction aborted. + */ +HTSV_Result +HeapTupleSatisfiesVacuumHorizon(HeapTuple htup, Buffer buffer, TransactionId *dead_after) { HeapTupleHeader tuple = htup->t_data; Assert(ItemPointerIsValid(&htup->t_self)); Assert(htup->t_tableOid != InvalidOid); + Assert(dead_after != NULL); + + *dead_after = InvalidTransactionId; /* * Has inserting transaction committed? @@ -1323,17 +1360,15 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, else if (TransactionIdDidCommit(xmax)) { /* - * The multixact might still be running due to lockers. If the - * updater is below the xid horizon, we have to return DEAD - * regardless -- otherwise we could end up with a tuple where the - * updater has to be removed due to the horizon, but is not pruned - * away. It's not a problem to prune that tuple, because any - * remaining lockers will also be present in newer tuple versions. + * The multixact might still be running due to lockers. Need to + * allow for pruning if below the xid horizon regardless -- + * otherwise we could end up with a tuple where the updater has to + * be removed due to the horizon, but is not pruned away. It's + * not a problem to prune that tuple, because any remaining + * lockers will also be present in newer tuple versions. */ - if (!TransactionIdPrecedes(xmax, OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - return HEAPTUPLE_DEAD; + *dead_after = xmax; + return HEAPTUPLE_RECENTLY_DEAD; } else if (!MultiXactIdIsRunning(HeapTupleHeaderGetRawXmax(tuple), false)) { @@ -1372,14 +1407,11 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin, } /* - * Deleter committed, but perhaps it was recent enough that some open - * transactions could still see the tuple. + * Deleter committed, allow caller to check if it was recent enough that + * some open transactions could still see the tuple. */ - if (!TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin)) - return HEAPTUPLE_RECENTLY_DEAD; - - /* Otherwise, it's dead and removable */ - return HEAPTUPLE_DEAD; + *dead_after = HeapTupleHeaderGetRawXmax(tuple); + return HEAPTUPLE_RECENTLY_DEAD; } @@ -1418,7 +1450,7 @@ HeapTupleSatisfiesNonVacuumable(HeapTuple htup, Snapshot snapshot, * if the tuple is removable. */ bool -HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) +HeapTupleIsSurelyDead(GlobalVisState *vistest, HeapTuple htup) { HeapTupleHeader tuple = htup->t_data; @@ -1459,7 +1491,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin) return false; /* Deleter committed, so tuple is dead if the XID is old enough. */ - return TransactionIdPrecedes(HeapTupleHeaderGetRawXmax(tuple), OldestXmin); + return GlobalVisTestIsRemovableXid(vistest, + HeapTupleHeaderGetRawXmax(tuple)); } /* diff --git a/src/backend/access/heap/pruneheap.c b/src/backend/access/heap/pruneheap.c index 256df4de105..00a3cb106aa 100644 --- a/src/backend/access/heap/pruneheap.c +++ b/src/backend/access/heap/pruneheap.c @@ -23,12 +23,30 @@ #include "miscadmin.h" #include "pgstat.h" #include "storage/bufmgr.h" +#include "utils/snapmgr.h" #include "utils/rel.h" #include "utils/snapmgr.h" /* Working data for heap_page_prune and subroutines */ typedef struct { + Relation rel; + + /* tuple visibility test, initialized for the relation */ + GlobalVisState *vistest; + + /* + * Thresholds set by TransactionIdLimitedForOldSnapshots() if they have + * been computed (done on demand, and only if + * OldSnapshotThresholdActive()). The first time a tuple is about to be + * removed based on the limited horizon, old_snap_used is set to true, and + * SetOldSnapshotThresholdTimestamp() is called. See + * heap_prune_satisfies_vacuum(). + */ + TimestampTz old_snap_ts; + TransactionId old_snap_xmin; + bool old_snap_used; + TransactionId new_prune_xid; /* new prune hint value for page */ TransactionId latestRemovedXid; /* latest xid to be removed by this prune */ int nredirected; /* numbers of entries in arrays below */ @@ -43,9 +61,8 @@ typedef struct } PruneState; /* Local functions */ -static int heap_prune_chain(Relation relation, Buffer buffer, +static int heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, PruneState *prstate); static void heap_prune_record_prunable(PruneState *prstate, TransactionId xid); static void heap_prune_record_redirect(PruneState *prstate, @@ -65,16 +82,16 @@ static void heap_prune_record_unused(PruneState *prstate, OffsetNumber offnum); * if there's not any use in pruning. * * Caller must have pin on the buffer, and must *not* have a lock on it. - * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). */ void heap_page_prune_opt(Relation relation, Buffer buffer) { Page page = BufferGetPage(buffer); + TransactionId prune_xid; + GlobalVisState *vistest; + TransactionId limited_xmin = InvalidTransactionId; + TimestampTz limited_ts = 0; Size minfree; - TransactionId OldestXmin; /* * We can't write WAL in recovery mode, so there's no point trying to @@ -85,37 +102,55 @@ heap_page_prune_opt(Relation relation, Buffer buffer) return; /* - * Use the appropriate xmin horizon for this relation. If it's a proper - * catalog relation or a user defined, additional, catalog relation, we - * need to use the horizon that includes slots, otherwise the data-only - * horizon can be used. Note that the toast relation of user defined - * relations are *not* considered catalog relations. + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ + if (old_snapshot_threshold == 0) + SnapshotTooOldMagicForTest(); + + /* + * First check whether there's any chance there's something to prune, + * determining the appropriate horizon is a waste if there's no prune_xid + * (i.e. no updates/deletes left potentially dead tuples around). + */ + prune_xid = ((PageHeader) page)->pd_prune_xid; + if (!TransactionIdIsValid(prune_xid)) + return; + + /* + * Check whether prune_xid indicates that there may be dead rows that can + * be cleaned up. * - * It is OK to apply the old snapshot limit before acquiring the cleanup + * It is OK to check the old snapshot limit before acquiring the cleanup * lock because the worst that can happen is that we are not quite as * aggressive about the cleanup (by however many transaction IDs are * consumed between this point and acquiring the lock). This allows us to * save significant overhead in the case where the page is found not to be * prunable. - */ - if (IsCatalogRelation(relation) || - RelationIsAccessibleInLogicalDecoding(relation)) - OldestXmin = RecentGlobalXmin; - else - OldestXmin = - TransactionIdLimitedForOldSnapshots(RecentGlobalDataXmin, - relation); - - Assert(TransactionIdIsValid(OldestXmin)); - - /* - * Let's see if we really need pruning. * - * Forget it if page is not hinted to contain something prunable that's - * older than OldestXmin. + * Even if old_snapshot_threshold is set, we first check whether the page + * can be pruned without. Both because + * TransactionIdLimitedForOldSnapshots() is not cheap, and because not + * unnecessarily relying on old_snapshot_threshold avoids causing + * conflicts. */ - if (!PageIsPrunable(page, OldestXmin)) - return; + vistest = GlobalVisTestFor(relation); + + if (!GlobalVisTestIsRemovableXid(vistest, prune_xid)) + { + if (!OldSnapshotThresholdActive()) + return; + + if (!TransactionIdLimitedForOldSnapshots(GlobalVisTestNonRemovableHorizon(vistest), + relation, + &limited_xmin, &limited_ts)) + return; + + if (!TransactionIdPrecedes(prune_xid, limited_xmin)) + return; + } /* * We prune when a previous UPDATE failed to find enough space on the page @@ -151,7 +186,9 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * needed */ /* OK to prune */ - (void) heap_page_prune(relation, buffer, OldestXmin, true, &ignore); + (void) heap_page_prune(relation, buffer, vistest, + limited_xmin, limited_ts, + true, &ignore); } /* And release buffer lock */ @@ -165,8 +202,11 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * * Caller must have pin and buffer cleanup lock on the page. * - * OldestXmin is the cutoff XID used to distinguish whether tuples are DEAD - * or RECENTLY_DEAD (see HeapTupleSatisfiesVacuum). + * vistest is used to distinguish whether tuples are DEAD or RECENTLY_DEAD + * (see heap_prune_satisfies_vacuum and + * HeapTupleSatisfiesVacuum). old_snap_xmin / old_snap_ts need to + * either have been set by TransactionIdLimitedForOldSnapshots, or + * InvalidTransactionId/0 respectively. * * If report_stats is true then we send the number of reclaimed heap-only * tuples to pgstats. (This must be false during vacuum, since vacuum will @@ -177,7 +217,10 @@ heap_page_prune_opt(Relation relation, Buffer buffer) * latestRemovedXid. */ int -heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, +heap_page_prune(Relation relation, Buffer buffer, + GlobalVisState *vistest, + TransactionId old_snap_xmin, + TimestampTz old_snap_ts, bool report_stats, TransactionId *latestRemovedXid) { int ndeleted = 0; @@ -198,6 +241,11 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * initialize the rest of our working state. */ prstate.new_prune_xid = InvalidTransactionId; + prstate.rel = relation; + prstate.vistest = vistest; + prstate.old_snap_xmin = old_snap_xmin; + prstate.old_snap_ts = old_snap_ts; + prstate.old_snap_used = false; prstate.latestRemovedXid = *latestRemovedXid; prstate.nredirected = prstate.ndead = prstate.nunused = 0; memset(prstate.marked, 0, sizeof(prstate.marked)); @@ -220,9 +268,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, continue; /* Process this item or chain of items */ - ndeleted += heap_prune_chain(relation, buffer, offnum, - OldestXmin, - &prstate); + ndeleted += heap_prune_chain(buffer, offnum, &prstate); } /* Any error while applying the changes is critical */ @@ -323,6 +369,85 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, } +/* + * Perform visiblity checks for heap pruning. + * + * This is more complicated than just using GlobalVisTestIsRemovableXid() + * because of old_snapshot_threshold. We only want to increase the threshold + * that triggers errors for old snapshots when we actually decide to remove a + * row based on the limited horizon. + * + * Due to its cost we also only want to call + * TransactionIdLimitedForOldSnapshots() if necessary, i.e. we might not have + * done so in heap_hot_prune_opt() if pd_prune_xid was old enough. But we + * still want to be able to remove rows that are too new to be removed + * according to prstate->vistest, but that can be removed based on + * old_snapshot_threshold. So we call TransactionIdLimitedForOldSnapshots() on + * demand in here, if appropriate. + */ +static HTSV_Result +heap_prune_satisfies_vacuum(PruneState *prstate, HeapTuple tup, Buffer buffer) +{ + HTSV_Result res; + TransactionId dead_after; + + res = HeapTupleSatisfiesVacuumHorizon(tup, buffer, &dead_after); + + if (res != HEAPTUPLE_RECENTLY_DEAD) + return res; + + /* + * If we are already relying on the limited xmin, there is no need to + * delay doing so anymore. + */ + if (prstate->old_snap_used) + { + Assert(TransactionIdIsValid(prstate->old_snap_xmin)); + + if (TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + res = HEAPTUPLE_DEAD; + return res; + } + + /* + * First check if GlobalVisTestIsRemovableXid() is sufficient to find the + * row dead. If not, and old_snapshot_threshold is enabled, try to use the + * lowered horizon. + */ + if (GlobalVisTestIsRemovableXid(prstate->vistest, dead_after)) + res = HEAPTUPLE_DEAD; + else if (OldSnapshotThresholdActive()) + { + /* haven't determined limited horizon yet, requests */ + if (!TransactionIdIsValid(prstate->old_snap_xmin)) + { + TransactionId horizon = + GlobalVisTestNonRemovableHorizon(prstate->vistest); + + TransactionIdLimitedForOldSnapshots(horizon, prstate->rel, + &prstate->old_snap_xmin, + &prstate->old_snap_ts); + } + + if (TransactionIdIsValid(prstate->old_snap_xmin) && + TransactionIdPrecedes(dead_after, prstate->old_snap_xmin)) + { + /* + * About to remove row based on snapshot_too_old. Need to raise + * the threshold so problematic accesses would error. + */ + Assert(!prstate->old_snap_used); + SetOldSnapshotThresholdTimestamp(prstate->old_snap_ts, + prstate->old_snap_xmin); + prstate->old_snap_used = true; + res = HEAPTUPLE_DEAD; + } + } + + return res; +} + + /* * Prune specified line pointer or a HOT chain originating at line pointer. * @@ -349,9 +474,7 @@ heap_page_prune(Relation relation, Buffer buffer, TransactionId OldestXmin, * Returns the number of tuples (to be) deleted from the page. */ static int -heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, - TransactionId OldestXmin, - PruneState *prstate) +heap_prune_chain(Buffer buffer, OffsetNumber rootoffnum, PruneState *prstate) { int ndeleted = 0; Page dp = (Page) BufferGetPage(buffer); @@ -366,7 +489,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, i; HeapTupleData tup; - tup.t_tableOid = RelationGetRelid(relation); + tup.t_tableOid = RelationGetRelid(prstate->rel); rootlp = PageGetItemId(dp, rootoffnum); @@ -401,7 +524,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, * either here or while following a chain below. Whichever path * gets there first will mark the tuple unused. */ - if (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer) + if (heap_prune_satisfies_vacuum(prstate, &tup, buffer) == HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup)) { heap_prune_record_unused(prstate, rootoffnum); @@ -485,7 +608,7 @@ heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum, */ tupdead = recent_dead = false; - switch (HeapTupleSatisfiesVacuum(&tup, OldestXmin, buffer)) + switch (heap_prune_satisfies_vacuum(prstate, &tup, buffer)) { case HEAPTUPLE_DEAD: tupdead = true; diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 1bbc4598f75..44e2224dd55 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -788,6 +788,7 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, PROGRESS_VACUUM_MAX_DEAD_TUPLES }; int64 initprog_val[3]; + GlobalVisState *vistest; pg_rusage_init(&ru0); @@ -816,6 +817,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, vacrelstats->nonempty_pages = 0; vacrelstats->latestRemovedXid = InvalidTransactionId; + vistest = GlobalVisTestFor(onerel); + /* * Initialize state for a parallel vacuum. As of now, only one worker can * be used for an index, so we invoke parallelism only if there are at @@ -1239,7 +1242,8 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, * * We count tuples removed by the pruning step as removed by VACUUM. */ - tups_vacuumed += heap_page_prune(onerel, buf, OldestXmin, false, + tups_vacuumed += heap_page_prune(onerel, buf, vistest, false, + InvalidTransactionId, 0, &vacrelstats->latestRemovedXid); /* @@ -1596,14 +1600,16 @@ lazy_scan_heap(Relation onerel, VacuumParams *params, LVRelStats *vacrelstats, } /* - * It's possible for the value returned by GetOldestXmin() to move - * backwards, so it's not wrong for us to see tuples that appear to - * not be visible to everyone yet, while PD_ALL_VISIBLE is already - * set. The real safe xmin value never moves backwards, but - * GetOldestXmin() is conservative and sometimes returns a value - * that's unnecessarily small, so if we see that contradiction it just - * means that the tuples that we think are not visible to everyone yet - * actually are, and the PD_ALL_VISIBLE flag is correct. + * It's possible for the value returned by + * GetOldestNonRemovableTransactionId() to move backwards, so it's not + * wrong for us to see tuples that appear to not be visible to + * everyone yet, while PD_ALL_VISIBLE is already set. The real safe + * xmin value never moves backwards, but + * GetOldestNonRemovableTransactionId() is conservative and sometimes + * returns a value that's unnecessarily small, so if we see that + * contradiction it just means that the tuples that we think are not + * visible to everyone yet actually are, and the PD_ALL_VISIBLE flag + * is correct. * * There should never be dead tuples on a page with PD_ALL_VISIBLE * set, however. diff --git a/src/backend/access/index/indexam.c b/src/backend/access/index/indexam.c index 6b9750c244a..3fb8688f8f4 100644 --- a/src/backend/access/index/indexam.c +++ b/src/backend/access/index/indexam.c @@ -519,7 +519,8 @@ index_getnext_tid(IndexScanDesc scan, ScanDirection direction) SCAN_CHECKS; CHECK_SCAN_PROCEDURE(amgettuple); - Assert(TransactionIdIsValid(RecentGlobalXmin)); + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(TransactionIdIsValid(RecentXmin)); /* * The AM's amgettuple proc finds the next index entry matching the scan diff --git a/src/backend/access/nbtree/README b/src/backend/access/nbtree/README index abce31a5a96..781a8f1932d 100644 --- a/src/backend/access/nbtree/README +++ b/src/backend/access/nbtree/README @@ -342,9 +342,9 @@ snapshots and registered snapshots as of the deletion are gone; which is overly strong, but is simple to implement within Postgres. When marked dead, a deleted page is labeled with the next-transaction counter value. VACUUM can reclaim the page for re-use when this transaction number is -older than RecentGlobalXmin. As collateral damage, this implementation -also waits for running XIDs with no snapshots and for snapshots taken -until the next transaction to allocate an XID commits. +guaranteed to be "visible to everyone". As collateral damage, this +implementation also waits for running XIDs with no snapshots and for +snapshots taken until the next transaction to allocate an XID commits. Reclaiming a page doesn't actually change its state on disk --- we simply record it in the shared-memory free space map, from which it will be @@ -411,8 +411,8 @@ page and also the correct place to hold the current value. We can avoid the cost of walking down the tree in such common cases. The optimization works on the assumption that there can only be one -non-ignorable leaf rightmost page, and so even a RecentGlobalXmin style -interlock isn't required. We cannot fail to detect that our hint was +non-ignorable leaf rightmost page, and so not even a visible-to-everyone +style interlock required. We cannot fail to detect that our hint was invalidated, because there can only be one such page in the B-Tree at any time. It's possible that the page will be deleted and recycled without a backend's cached page also being detected as invalidated, but diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c index d5db9aaa3a1..74be3807bb7 100644 --- a/src/backend/access/nbtree/nbtpage.c +++ b/src/backend/access/nbtree/nbtpage.c @@ -1097,7 +1097,7 @@ _bt_page_recyclable(Page page) */ opaque = (BTPageOpaque) PageGetSpecialPointer(page); if (P_ISDELETED(opaque) && - TransactionIdPrecedes(opaque->btpo.xact, RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, opaque->btpo.xact)) return true; return false; } @@ -2318,7 +2318,7 @@ _bt_unlink_halfdead_page(Relation rel, Buffer leafbuf, BlockNumber scanblkno, * updated links to the target, ReadNewTransactionId() suffices as an * upper bound. Any scan having retained a now-stale link is advertising * in its PGXACT an xmin less than or equal to the value we read here. It - * will continue to do so, holding back RecentGlobalXmin, for the duration + * will continue to do so, holding back the xmin horizon, for the duration * of that scan. */ page = BufferGetPage(buf); diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 49a8a9708e3..8fa6ac7296b 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -808,6 +808,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) metapg = BufferGetPage(metabuf); metad = BTPageGetMeta(metapg); + /* + * XXX: If IndexVacuumInfo contained the heap relation, we could be more + * aggressive about vacuuming non catalog relations by passing the table + * to GlobalVisCheckRemovableXid(). + */ + if (metad->btm_version < BTREE_NOVAC_VERSION) { /* @@ -817,13 +823,12 @@ _bt_vacuum_needs_cleanup(IndexVacuumInfo *info) result = true; } else if (TransactionIdIsValid(metad->btm_oldest_btpo_xact) && - TransactionIdPrecedes(metad->btm_oldest_btpo_xact, - RecentGlobalXmin)) + GlobalVisCheckRemovableXid(NULL, metad->btm_oldest_btpo_xact)) { /* * If any oldest btpo.xact from a previously deleted page in the index - * is older than RecentGlobalXmin, then at least one deleted page can - * be recycled -- don't skip cleanup. + * is visible to everyone, then at least one deleted page can be + * recycled -- don't skip cleanup. */ result = true; } @@ -1276,14 +1281,13 @@ backtrack: * own conflict now.) * * Backends with snapshots acquired after a VACUUM starts but - * before it finishes could have a RecentGlobalXmin with a - * later xid than the VACUUM's OldestXmin cutoff. These - * backends might happen to opportunistically mark some index - * tuples LP_DEAD before we reach them, even though they may - * be after our cutoff. We don't try to kill these "extra" - * index tuples in _bt_delitems_vacuum(). This keep things - * simple, and allows us to always avoid generating our own - * conflicts. + * before it finishes could have visibility cutoff with a + * later xid than VACUUM's OldestXmin cutoff. These backends + * might happen to opportunistically mark some index tuples + * LP_DEAD before we reach them, even though they may be after + * our cutoff. We don't try to kill these "extra" index + * tuples in _bt_delitems_vacuum(). This keep things simple, + * and allows us to always avoid generating our own conflicts. */ Assert(!BTreeTupleIsPivot(itup)); if (!BTreeTupleIsPosting(itup)) diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c index dbec58d5249..bda9be23489 100644 --- a/src/backend/access/nbtree/nbtxlog.c +++ b/src/backend/access/nbtree/nbtxlog.c @@ -948,11 +948,11 @@ btree_xlog_reuse_page(XLogReaderState *record) * Btree reuse_page records exist to provide a conflict point when we * reuse pages in the index via the FSM. That's all they do though. * - * latestRemovedXid was the page's btpo.xact. The btpo.xact < - * RecentGlobalXmin test in _bt_page_recyclable() conceptually mirrors the - * pgxact->xmin > limitXmin test in GetConflictingVirtualXIDs(). - * Consequently, one XID value achieves the same exclusion effect on - * primary and standby. + * latestRemovedXid was the page's btpo.xact. The + * GlobalVisCheckRemovableXid test in _bt_page_recyclable() conceptually + * mirrors the pgxact->xmin > limitXmin test in + * GetConflictingVirtualXIDs(). Consequently, one XID value achieves the + * same exclusion effect on primary and standby. */ if (InHotStandby) { diff --git a/src/backend/access/spgist/spgvacuum.c b/src/backend/access/spgist/spgvacuum.c index bd98707f3c0..e1c58933f97 100644 --- a/src/backend/access/spgist/spgvacuum.c +++ b/src/backend/access/spgist/spgvacuum.c @@ -501,10 +501,14 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) OffsetNumber itemToPlaceholder[MaxIndexTuplesPerPage]; OffsetNumber itemnos[MaxIndexTuplesPerPage]; spgxlogVacuumRedirect xlrec; + GlobalVisState *vistest; xlrec.nToPlaceholder = 0; xlrec.newestRedirectXid = InvalidTransactionId; + /* XXX: providing heap relation would allow more pruning */ + vistest = GlobalVisTestFor(NULL); + START_CRIT_SECTION(); /* @@ -521,7 +525,7 @@ vacuumRedirectAndPlaceholder(Relation index, Buffer buffer) dt = (SpGistDeadTuple) PageGetItem(page, PageGetItemId(page, i)); if (dt->tupstate == SPGIST_REDIRECT && - TransactionIdPrecedes(dt->xid, RecentGlobalXmin)) + GlobalVisTestIsRemovableXid(vistest, dt->xid)) { dt->tupstate = SPGIST_PLACEHOLDER; Assert(opaque->nRedirection > 0); diff --git a/src/backend/access/transam/README b/src/backend/access/transam/README index eb9aac5fd39..fffe0783295 100644 --- a/src/backend/access/transam/README +++ b/src/backend/access/transam/README @@ -293,42 +293,50 @@ once, rather than assume they can read it multiple times and get the same answer each time. (Use volatile-qualified pointers when doing this, to ensure that the C compiler does exactly what you tell it to.) -Another important activity that uses the shared ProcArray is GetOldestXmin, -which must determine a lower bound for the oldest xmin of any active MVCC -snapshot, system-wide. Each individual backend advertises the smallest -xmin of its own snapshots in MyPgXact->xmin, or zero if it currently has no -live snapshots (eg, if it's between transactions or hasn't yet set a -snapshot for a new transaction). GetOldestXmin takes the MIN() of the -valid xmin fields. It does this with only shared lock on ProcArrayLock, -which means there is a potential race condition against other backends -doing GetSnapshotData concurrently: we must be certain that a concurrent -backend that is about to set its xmin does not compute an xmin less than -what GetOldestXmin returns. We ensure that by including all the active -XIDs into the MIN() calculation, along with the valid xmins. The rule that -transactions can't exit without taking exclusive ProcArrayLock ensures that -concurrent holders of shared ProcArrayLock will compute the same minimum of -currently-active XIDs: no xact, in particular not the oldest, can exit -while we hold shared ProcArrayLock. So GetOldestXmin's view of the minimum -active XID will be the same as that of any concurrent GetSnapshotData, and -so it can't produce an overestimate. If there is no active transaction at -all, GetOldestXmin returns latestCompletedXid + 1, which is a lower bound -for the xmin that might be computed by concurrent or later GetSnapshotData -calls. (We know that no XID less than this could be about to appear in -the ProcArray, because of the XidGenLock interlock discussed above.) +Another important activity that uses the shared ProcArray is +ComputeXidHorizons, which must determine a lower bound for the oldest xmin +of any active MVCC snapshot, system-wide. Each individual backend +advertises the smallest xmin of its own snapshots in MyPgXact->xmin, or zero +if it currently has no live snapshots (eg, if it's between transactions or +hasn't yet set a snapshot for a new transaction). ComputeXidHorizons takes +the MIN() of the valid xmin fields. It does this with only shared lock on +ProcArrayLock, which means there is a potential race condition against other +backends doing GetSnapshotData concurrently: we must be certain that a +concurrent backend that is about to set its xmin does not compute an xmin +less than what ComputeXidHorizons determines. We ensure that by including +all the active XIDs into the MIN() calculation, along with the valid xmins. +The rule that transactions can't exit without taking exclusive ProcArrayLock +ensures that concurrent holders of shared ProcArrayLock will compute the +same minimum of currently-active XIDs: no xact, in particular not the +oldest, can exit while we hold shared ProcArrayLock. So +ComputeXidHorizons's view of the minimum active XID will be the same as that +of any concurrent GetSnapshotData, and so it can't produce an overestimate. +If there is no active transaction at all, ComputeXidHorizons uses +latestCompletedXid + 1, which is a lower bound for the xmin that might +be computed by concurrent or later GetSnapshotData calls. (We know that no +XID less than this could be about to appear in the ProcArray, because of the +XidGenLock interlock discussed above.) -GetSnapshotData also performs an oldest-xmin calculation (which had better -match GetOldestXmin's) and stores that into RecentGlobalXmin, which is used -for some tuple age cutoff checks where a fresh call of GetOldestXmin seems -too expensive. Note that while it is certain that two concurrent -executions of GetSnapshotData will compute the same xmin for their own -snapshots, as argued above, it is not certain that they will arrive at the -same estimate of RecentGlobalXmin. This is because we allow XID-less -transactions to clear their MyPgXact->xmin asynchronously (without taking -ProcArrayLock), so one execution might see what had been the oldest xmin, -and another not. This is OK since RecentGlobalXmin need only be a valid -lower bound. As noted above, we are already assuming that fetch/store -of the xid fields is atomic, so assuming it for xmin as well is no extra -risk. +As GetSnapshotData is performance critical, it does not perform an accurate +oldest-xmin calculation (it used to, until v13). The contents of a snapshot +only depend on the xids of other backends, not their xmin. As backend's xmin +changes much more often than its xid, having GetSnapshotData look at xmins +can lead to a lot of unnecessary cacheline ping-pong. Instead +GetSnapshotData updates approximate thresholds (one that guarantees that all +deleted rows older than it can be removed, another determining that deleted +rows newer than it can not be removed). GlobalVisTest* uses those threshold +to make invisibility decision, falling back to ComputeXidHorizons if +necessary. + +Note that while it is certain that two concurrent executions of +GetSnapshotData will compute the same xmin for their own snapshots, there is +no such guarantee for the horizons computed by ComputeXidHorizons. This is +because we allow XID-less transactions to clear their MyPgXact->xmin +asynchronously (without taking ProcArrayLock), so one execution might see +what had been the oldest xmin, and another not. This is OK since the +thresholds need only be a valid lower bound. As noted above, we are already +assuming that fetch/store of the xid fields is atomic, so assuming it for +xmin as well is no extra risk. pg_xact and pg_subtrans diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 8f72faee82c..09c01ed4ae4 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9096,7 +9096,7 @@ CreateCheckPoint(int flags) * StartupSUBTRANS hasn't been called yet. */ if (!RecoveryInProgress()) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update stats before releasing lock. */ LogCheckpointEnd(false); @@ -9456,7 +9456,7 @@ CreateRestartPoint(int flags) * this because StartupSUBTRANS hasn't been called yet. */ if (EnableHotStandby) - TruncateSUBTRANS(GetOldestXmin(NULL, PROCARRAY_FLAGS_DEFAULT)); + TruncateSUBTRANS(GetOldestTransactionIdConsideredRunning()); /* Real work is done, but log and update before releasing lock. */ LogCheckpointEnd(true); diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index e0fa73ba790..8af12b5c6b2 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -1045,7 +1045,7 @@ acquire_sample_rows(Relation onerel, int elevel, totalblocks = RelationGetNumberOfBlocks(onerel); /* Need a cutoff xmin for HeapTupleSatisfiesVacuum */ - OldestXmin = GetOldestXmin(onerel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(onerel); /* Prepare for sampling block numbers */ nblocks = BlockSampler_Init(&bs, totalblocks, targrows, random()); diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 576c7e63e99..22228f5684f 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -955,8 +955,25 @@ vacuum_set_xid_limits(Relation rel, * working on a particular table at any time, and that each vacuum is * always an independent transaction. */ - *oldestXmin = - TransactionIdLimitedForOldSnapshots(GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM), rel); + *oldestXmin = GetOldestNonRemovableTransactionId(rel); + + if (OldSnapshotThresholdActive()) + { + TransactionId limit_xmin; + TimestampTz limit_ts; + + if (TransactionIdLimitedForOldSnapshots(*oldestXmin, rel, &limit_xmin, &limit_ts)) + { + /* + * TODO: We should only set the threshold if we are pruning on the + * basis of the increased limits. Not as crucial here as it is for + * opportunistic pruning (which often happens at a much higher + * frequency), but would still be a significant improvement. + */ + SetOldSnapshotThresholdTimestamp(limit_ts, limit_xmin); + *oldestXmin = limit_xmin; + } + } Assert(TransactionIdIsNormal(*oldestXmin)); @@ -1345,12 +1362,13 @@ vac_update_datfrozenxid(void) bool dirty = false; /* - * Initialize the "min" calculation with GetOldestXmin, which is a - * reasonable approximation to the minimum relfrozenxid for not-yet- - * committed pg_class entries for new tables; see AddNewRelationTuple(). - * So we cannot produce a wrong minimum by starting with this. + * Initialize the "min" calculation with + * GetOldestNonRemovableTransactionId(), which is a reasonable + * approximation to the minimum relfrozenxid for not-yet-committed + * pg_class entries for new tables; see AddNewRelationTuple(). So we + * cannot produce a wrong minimum by starting with this. */ - newFrozenXid = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + newFrozenXid = GetOldestNonRemovableTransactionId(NULL); /* * Similarly, initialize the MultiXact "min" with the value that would be @@ -1681,8 +1699,9 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) StartTransactionCommand(); /* - * Functions in indexes may want a snapshot set. Also, setting a snapshot - * ensures that RecentGlobalXmin is kept truly recent. + * Need to acquire a snapshot to prevent pg_subtrans from being truncated, + * cutoff xids in local memory wrapping around, and to have updated xmin + * horizons. */ PushActiveSnapshot(GetTransactionSnapshot()); @@ -1705,8 +1724,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params) * * Note: these flags remain set until CommitTransaction or * AbortTransaction. We don't want to clear them until we reset - * MyPgXact->xid/xmin, else OldestXmin might appear to go backwards, - * which is probably Not Good. + * MyPgXact->xid/xmin, otherwise GetOldestNonRemovableTransactionId() + * might appear to go backwards, which is probably Not Good. */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyPgXact->vacuumFlags |= PROC_IN_VACUUM; diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 9c7d4b0c60e..ac97e28be19 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -1877,6 +1877,10 @@ get_database_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index ff985b9b24c..bdaf0312d63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -122,6 +122,10 @@ get_subscription_list(void) * the secondary effect that it sets RecentGlobalXmin. (This is critical * for anything that reads heap pages, because HOT may decide to prune * them even if the process doesn't attempt to modify any tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ StartTransactionCommand(); (void) GetTransactionSnapshot(); diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index d5a9b568a68..7c11e1ab44c 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -1181,22 +1181,7 @@ XLogWalRcvSendHSFeedback(bool immed) */ if (hot_standby_feedback) { - TransactionId slot_xmin; - - /* - * Usually GetOldestXmin() would include both global replication slot - * xmin and catalog_xmin in its calculations, but we want to derive - * separate values for each of those. So we ask for an xmin that - * excludes the catalog_xmin. - */ - xmin = GetOldestXmin(NULL, - PROCARRAY_FLAGS_DEFAULT | PROCARRAY_SLOTS_XMIN); - - ProcArrayGetReplicationSlotXmin(&slot_xmin, &catalog_xmin); - - if (TransactionIdIsValid(slot_xmin) && - TransactionIdPrecedes(slot_xmin, xmin)) - xmin = slot_xmin; + GetReplicationHorizons(&xmin, &catalog_xmin); } else { diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index d13220c1400..460ca3f947f 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -2113,9 +2113,10 @@ ProcessStandbyHSFeedbackMessage(void) /* * Set the WalSender's xmin equal to the standby's requested xmin, so that - * the xmin will be taken into account by GetOldestXmin. This will hold - * back the removal of dead rows and thereby prevent the generation of - * cleanup conflicts on the standby server. + * the xmin will be taken into account by GetSnapshotData() / + * ComputeXidHorizons(). This will hold back the removal of dead rows and + * thereby prevent the generation of cleanup conflicts on the standby + * server. * * There is a small window for a race condition here: although we just * checked that feedbackXmin precedes nextXid, the nextXid could have @@ -2128,10 +2129,10 @@ ProcessStandbyHSFeedbackMessage(void) * own xmin would prevent nextXid from advancing so far. * * We don't bother taking the ProcArrayLock here. Setting the xmin field - * is assumed atomic, and there's no real need to prevent a concurrent - * GetOldestXmin. (If we're moving our xmin forward, this is obviously - * safe, and if we're moving it backwards, well, the data is at risk - * already since a VACUUM could have just finished calling GetOldestXmin.) + * is assumed atomic, and there's no real need to prevent concurrent + * horizon determinations. (If we're moving our xmin forward, this is + * obviously safe, and if we're moving it backwards, well, the data is at + * risk already since a VACUUM could already have determined the horizon.) * * If we're using a replication slot we reserve the xmin via that, * otherwise via the walsender's PGXACT entry. We can only track the diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index 522518695ee..360e6e9da07 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -99,6 +99,142 @@ typedef struct ProcArrayStruct int pgprocnos[FLEXIBLE_ARRAY_MEMBER]; } ProcArrayStruct; +/* + * State for the GlobalVisTest* family of functions. Those functions can + * e.g. be used to decide if a deleted row can be removed without violating + * MVCC semantics: If the deleted row's xmax is not considered to be running + * by anyone, the row can be removed. + * + * To avoid slowing down GetSnapshotData(), we don't calculate a precise + * cutoff XID while building a snapshot (looking at the frequently changing + * xmins scales badly). Instead we compute two boundaries while building the + * snapshot: + * + * 1) definitely_needed, indicating that rows deleted by XIDs >= + * definitely_needed are definitely still visible. + * + * 2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can + * definitely be removed + * + * When testing an XID that falls in between the two (i.e. XID >= maybe_needed + * && XID < definitely_needed), the boundaries can be recomputed (using + * ComputeXidHorizons()) to get a more accurate answer. This is cheaper than + * maintaining an accurate value all the time. + * + * As it is not cheap to compute accurate boundaries, we limit the number of + * times that happens in short succession. See GlobalVisTestShouldUpdate(). + * + * + * There are three backend lifetime instances of this struct, optimized for + * different types of relations. As e.g. a normal user defined table in one + * database is inaccessible to backends connected to another database, a test + * specific to a relation can be more aggressive than a test for a shared + * relation. Currently we track three different states: + * + * 1) GlobalVisSharedRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in any database, nor a + * replication slot's xmin, nor a replication slot's catalog_xmin might + * still consider XID as running. + * + * 2) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin, nor a replication slot's + * catalog_xmin might still consider XID as running. + * + * I.e. the difference to GlobalVisSharedRels is that + * snapshot in other databases are ignored. + * + * 3) GlobalVisCatalogRels, which only considers an XID's + * effects visible-to-everyone if neither snapshots in the current + * database, nor a replication slot's xmin consider XID as running. + * + * I.e. the difference to GlobalVisCatalogRels is that + * replication slot's catalog_xmin is not taken into account. + * + * GlobalVisTestFor(relation) returns the appropriate state + * for the relation. + * + * The boundaries are FullTransactionIds instead of TransactionIds to avoid + * wraparound dangers. There e.g. would otherwise exist no procarray state to + * prevent maybe_needed to become old enough after the GetSnapshotData() + * call. + * + * The typedef is in the header. + */ +struct GlobalVisState +{ + /* XIDs >= are considered running by some backend */ + FullTransactionId definitely_needed; + + /* XIDs < are not considered to be running by any backend */ + FullTransactionId maybe_needed; +}; + +/* + * Result of ComputeXidHorizons(). + */ +typedef struct ComputeXidHorizonsResult +{ + /* + * The value of ShmemVariableCache->latestCompletedXid when + * ComputeXidHorizons() held ProcArrayLock. + */ + FullTransactionId latest_completed; + + /* + * The same for procArray->replication_slot_xmin and. + * procArray->replication_slot_catalog_xmin. + */ + TransactionId slot_xmin; + TransactionId slot_catalog_xmin; + + /* + * Oldest xid that any backend might still consider running. This needs to + * include processes running VACUUM, in contrast to the normal visibility + * cutoffs, as vacuum needs to be able to perform pg_subtrans lookups when + * determining visibility, but doesn't care about rows above its xmin to + * be removed. + * + * This likely should only be needed to determine whether pg_subtrans can + * be truncated. It currently includes the effects of replications slots, + * for historical reasons. But that could likely be changed. + */ + TransactionId oldest_considered_running; + + /* + * Oldest xid for which deleted tuples need to be retained in shared + * tables. + * + * This includes the effects of replications lots. If that's not desired, + * look at shared_oldest_nonremovable_raw; + */ + TransactionId shared_oldest_nonremovable; + + /* + * Oldest xid that may be necessary to retain in shared tables. This is + * the same as shared_oldest_nonremovable, except that is not affected by + * replication slot's catalog_xmin. + * + * This is mainly useful to be able to send the catalog_xmin to upstream + * streaming replication servers via hot_standby_feedback, so they can + * apply the limit only when accessing catalog tables. + */ + TransactionId shared_oldest_nonremovable_raw; + + /* + * Oldest xid for which deleted tuples need to be retained in non-shared + * catalog tables. + */ + TransactionId catalog_oldest_nonremovable; + + /* + * Oldest xid for which deleted tuples need to be retained in normal user + * defined tables. + */ + TransactionId data_oldest_nonremovable; +} ComputeXidHorizonsResult; + + static ProcArrayStruct *procArray; static PGPROC *allProcs; @@ -118,6 +254,22 @@ static TransactionId latestObservedXid = InvalidTransactionId; */ static TransactionId standbySnapshotPendingXmin; +/* + * State for visibility checks on different types of relations. See struct + * GlobalVisState for details. As shared, catalog, and user defined + * relations can have different horizons, one such state exists for each. + */ +static GlobalVisState GlobalVisSharedRels; +static GlobalVisState GlobalVisCatalogRels; +static GlobalVisState GlobalVisDataRels; + +/* + * This backend's RecentXmin at the last time the accurate xmin horizon was + * recomputed, or InvalidTransactionId if it has not. Used to limit how many + * times accurate horizons are recomputed. See GlobalVisTestShouldUpdate(). + */ +static TransactionId ComputeXidHorizonsResultLastXmin; + #ifdef XIDCACHE_DEBUG /* counters for XidCache measurement */ @@ -180,6 +332,7 @@ static void MaintainLatestCompletedXidRecovery(TransactionId latestXid); static inline FullTransactionId FullXidRelativeTo(FullTransactionId rel, TransactionId xid); +static void GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons); /* * Report shared-memory space needed by CreateSharedProcArray. @@ -1302,159 +1455,191 @@ TransactionIdIsActive(TransactionId xid) /* - * GetOldestXmin -- returns oldest transaction that was running - * when any current transaction was started. + * Determine XID horizons. * - * If rel is NULL or a shared relation, all backends are considered, otherwise - * only backends running in this database are considered. + * This is used by wrapper functions like GetOldestNonRemovableTransactionId() + * (for VACUUM), GetReplicationHorizons() (for hot_standby_feedback), etc as + * well as "internally" by GlobalVisUpdate() (see comment above struct + * GlobalVisState). * - * The flags are used to ignore the backends in calculation when any of the - * corresponding flags is set. Typically, if you want to ignore ones with - * PROC_IN_VACUUM flag, you can use PROCARRAY_FLAGS_VACUUM. + * See the definition of ComputedXidHorizonsResult for the various computed + * horizons. * - * PROCARRAY_SLOTS_XMIN causes GetOldestXmin to ignore the xmin and - * catalog_xmin of any replication slots that exist in the system when - * calculating the oldest xmin. + * For VACUUM separate horizons (used to to decide which deleted tuples must + * be preserved), for shared and non-shared tables are computed. For shared + * relations backends in all databases must be considered, but for non-shared + * relations that's not required, since only backends in my own database could + * ever see the tuples in them. Also, we can ignore concurrently running lazy + * VACUUMs because (a) they must be working on other tables, and (b) they + * don't need to do snapshot-based lookups. * - * This is used by VACUUM to decide which deleted tuples must be preserved in - * the passed in table. For shared relations backends in all databases must be - * considered, but for non-shared relations that's not required, since only - * backends in my own database could ever see the tuples in them. Also, we can - * ignore concurrently running lazy VACUUMs because (a) they must be working - * on other tables, and (b) they don't need to do snapshot-based lookups. - * - * This is also used to determine where to truncate pg_subtrans. For that - * backends in all databases have to be considered, so rel = NULL has to be - * passed in. + * This also computes a horizon used to truncate pg_subtrans. For that + * backends in all databases have to be considered, and concurrently running + * lazy VACUUMs cannot be ignored, as they still may perform pg_subtrans + * accesses. * * Note: we include all currently running xids in the set of considered xids. * This ensures that if a just-started xact has not yet set its snapshot, * when it does set the snapshot it cannot set xmin less than what we compute. * See notes in src/backend/access/transam/README. * - * Note: despite the above, it's possible for the calculated value to move - * backwards on repeated calls. The calculated value is conservative, so that - * anything older is definitely not considered as running by anyone anymore, - * but the exact value calculated depends on a number of things. For example, - * if rel = NULL and there are no transactions running in the current - * database, GetOldestXmin() returns latestCompletedXid. If a transaction + * Note: despite the above, it's possible for the calculated values to move + * backwards on repeated calls. The calculated values are conservative, so + * that anything older is definitely not considered as running by anyone + * anymore, but the exact values calculated depend on a number of things. For + * example, if there are no transactions running in the current database, the + * horizon for normal tables will be latestCompletedXid. If a transaction * begins after that, its xmin will include in-progress transactions in other * databases that started earlier, so another call will return a lower value. * Nonetheless it is safe to vacuum a table in the current database with the * first result. There are also replication-related effects: a walsender * process can set its xmin based on transactions that are no longer running * on the primary but are still being replayed on the standby, thus possibly - * making the GetOldestXmin reading go backwards. In this case there is a - * possibility that we lose data that the standby would like to have, but - * unless the standby uses a replication slot to make its xmin persistent - * there is little we can do about that --- data is only protected if the - * walsender runs continuously while queries are executed on the standby. - * (The Hot Standby code deals with such cases by failing standby queries - * that needed to access already-removed data, so there's no integrity bug.) - * The return value is also adjusted with vacuum_defer_cleanup_age, so - * increasing that setting on the fly is another easy way to make - * GetOldestXmin() move backwards, with no consequences for data integrity. + * making the values go backwards. In this case there is a possibility that + * we lose data that the standby would like to have, but unless the standby + * uses a replication slot to make its xmin persistent there is little we can + * do about that --- data is only protected if the walsender runs continuously + * while queries are executed on the standby. (The Hot Standby code deals + * with such cases by failing standby queries that needed to access + * already-removed data, so there's no integrity bug.) The computed values + * are also adjusted with vacuum_defer_cleanup_age, so increasing that setting + * on the fly is another easy way to make horizons move backwards, with no + * consequences for data integrity. + * + * Note: the approximate horizons (see definition of GlobalVisState) are + * updated by the computations done here. That's currently required for + * correctness and a small optimization. Without doing so it's possible that + * heap vacuum's call to heap_page_prune() uses a more conservative horizon + * than later when deciding which tuples can be removed - which the code + * doesn't expect (breaking HOT). */ -TransactionId -GetOldestXmin(Relation rel, int flags) +static void +ComputeXidHorizons(ComputeXidHorizonsResult *h) { ProcArrayStruct *arrayP = procArray; - TransactionId result; - int index; - bool allDbs; + TransactionId kaxmin; + bool in_recovery = RecoveryInProgress(); - TransactionId replication_slot_xmin = InvalidTransactionId; - TransactionId replication_slot_catalog_xmin = InvalidTransactionId; - - /* - * If we're not computing a relation specific limit, or if a shared - * relation has been passed in, backends in all databases have to be - * considered. - */ - allDbs = rel == NULL || rel->rd_rel->relisshared; - - /* Cannot look for individual databases during recovery */ - Assert(allDbs || !RecoveryInProgress()); + /* inferred after ProcArrayLock is released */ + h->catalog_oldest_nonremovable = InvalidTransactionId; LWLockAcquire(ProcArrayLock, LW_SHARED); + h->latest_completed = ShmemVariableCache->latestCompletedXid; + /* * We initialize the MIN() calculation with latestCompletedXid + 1. This * is a lower bound for the XIDs that might appear in the ProcArray later, * and so protects us against overestimating the result due to future * additions. */ - result = XidFromFullTransactionId(ShmemVariableCache->latestCompletedXid); - TransactionIdAdvance(result); - Assert(TransactionIdIsNormal(result)); + { + TransactionId initial; - for (index = 0; index < arrayP->numProcs; index++) + initial = XidFromFullTransactionId(h->latest_completed); + Assert(TransactionIdIsValid(initial)); + TransactionIdAdvance(initial); + + h->oldest_considered_running = initial; + h->shared_oldest_nonremovable = initial; + h->data_oldest_nonremovable = initial; + } + + /* + * Fetch slot horizons while ProcArrayLock is held - the + * LWLockAcquire/LWLockRelease are a barrier, ensuring this happens inside + * the lock. + */ + h->slot_xmin = procArray->replication_slot_xmin; + h->slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + + for (int index = 0; index < arrayP->numProcs; index++) { int pgprocno = arrayP->pgprocnos[index]; PGPROC *proc = &allProcs[pgprocno]; PGXACT *pgxact = &allPgXact[pgprocno]; + TransactionId xid; + TransactionId xmin; - if (pgxact->vacuumFlags & (flags & PROCARRAY_PROC_FLAGS_MASK)) + /* Fetch xid just once - see GetNewTransactionId */ + xid = UINT32_ACCESS_ONCE(pgxact->xid); + xmin = UINT32_ACCESS_ONCE(pgxact->xmin); + + /* + * Consider both the transaction's Xmin, and its Xid. + * + * We must check both because a transaction might have an Xmin but not + * (yet) an Xid; conversely, if it has an Xid, that could determine + * some not-yet-set Xmin. + */ + xmin = TransactionIdOlder(xmin, xid); + + /* if neither is set, this proc doesn't influence the horizon */ + if (!TransactionIdIsValid(xmin)) continue; - if (allDbs || + /* + * Don't ignore any procs when determining which transactions might be + * considered running. While slots should ensure logical decoding + * backends are protected even without this check, it can't hurt to + * include them here as well.. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, xmin); + + /* + * Skip over backends either vacuuming (which is ok with rows being + * removed, as long as pg_subtrans is not truncated) or doing logical + * decoding (which manages xmin separately, check below). + */ + if (pgxact->vacuumFlags & (PROC_IN_VACUUM | PROC_IN_LOGICAL_DECODING)) + continue; + + /* shared tables need to take backends in all database into account */ + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, xmin); + + /* + * Normally queries in other databases are ignored for anything but + * the shared horizon. But in recovery we cannot compute an accurate + * per-database horizon as all xids are managed via the + * KnownAssignedXids machinery. + */ + if (in_recovery || proc->databaseId == MyDatabaseId || proc->databaseId == 0) /* always include WalSender */ { - /* Fetch xid just once - see GetNewTransactionId */ - TransactionId xid = UINT32_ACCESS_ONCE(pgxact->xid); - - /* First consider the transaction's own Xid, if any */ - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; - - /* - * Also consider the transaction's Xmin, if set. - * - * We must check both Xid and Xmin because a transaction might - * have an Xmin but not (yet) an Xid; conversely, if it has an - * Xid, that could determine some not-yet-set Xmin. - */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - TransactionIdPrecedes(xid, result)) - result = xid; + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, xmin); } } /* - * Fetch into local variable while ProcArrayLock is held - the - * LWLockRelease below is a barrier, ensuring this happens inside the - * lock. + * If in recovery fetch oldest xid in KnownAssignedXids, will be applied + * after lock is released. */ - replication_slot_xmin = procArray->replication_slot_xmin; - replication_slot_catalog_xmin = procArray->replication_slot_catalog_xmin; + if (in_recovery) + kaxmin = KnownAssignedXidsGetOldestXmin(); - if (RecoveryInProgress()) + /* + * No other information needed, so release the lock immediately. The rest + * of the computations can be done without a lock. + */ + LWLockRelease(ProcArrayLock); + + if (in_recovery) { - /* - * Check to see whether KnownAssignedXids contains an xid value older - * than the main procarray. - */ - TransactionId kaxmin = KnownAssignedXidsGetOldestXmin(); - - LWLockRelease(ProcArrayLock); - - if (TransactionIdIsNormal(kaxmin) && - TransactionIdPrecedes(kaxmin, result)) - result = kaxmin; + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, kaxmin); + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, kaxmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, kaxmin); } else { /* - * No other information needed, so release the lock immediately. - */ - LWLockRelease(ProcArrayLock); - - /* - * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age, - * being careful not to generate a "permanent" XID. + * Compute the cutoff XID by subtracting vacuum_defer_cleanup_age. * * vacuum_defer_cleanup_age provides some additional "slop" for the * benefit of hot standby queries on standby servers. This is quick @@ -1466,34 +1651,146 @@ GetOldestXmin(Relation rel, int flags) * in varsup.c. Also note that we intentionally don't apply * vacuum_defer_cleanup_age on standby servers. */ - result -= vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(result)) - result = FirstNormalTransactionId; + h->oldest_considered_running = + TransactionIdRetreatedBy(h->oldest_considered_running, + vacuum_defer_cleanup_age); + h->shared_oldest_nonremovable = + TransactionIdRetreatedBy(h->shared_oldest_nonremovable, + vacuum_defer_cleanup_age); + h->data_oldest_nonremovable = + TransactionIdRetreatedBy(h->data_oldest_nonremovable, + vacuum_defer_cleanup_age); } /* * Check whether there are replication slots requiring an older xmin. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, result)) - result = replication_slot_xmin; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, h->slot_xmin); + h->data_oldest_nonremovable = + TransactionIdOlder(h->data_oldest_nonremovable, h->slot_xmin); /* - * After locks have been released and vacuum_defer_cleanup_age has been - * applied, check whether we need to back up further to make logical - * decoding possible. We need to do so if we're computing the global limit - * (rel = NULL) or if the passed relation is a catalog relation of some - * kind. + * The only difference between catalog / data horizons is that the slot's + * catalog xmin is applied to the catalog one (so catalogs can be accessed + * for logical decoding). Initialize with data horizon, and then back up + * further if necessary. Have to back up the shared horizon as well, since + * that also can contain catalogs. */ - if (!(flags & PROCARRAY_SLOTS_XMIN) && - (rel == NULL || - RelationIsAccessibleInLogicalDecoding(rel)) && - TransactionIdIsValid(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, result)) - result = replication_slot_catalog_xmin; + h->shared_oldest_nonremovable_raw = h->shared_oldest_nonremovable; + h->shared_oldest_nonremovable = + TransactionIdOlder(h->shared_oldest_nonremovable, + h->slot_catalog_xmin); + h->catalog_oldest_nonremovable = h->data_oldest_nonremovable; + h->catalog_oldest_nonremovable = + TransactionIdOlder(h->catalog_oldest_nonremovable, + h->slot_catalog_xmin); - return result; + /* + * It's possible that slots / vacuum_defer_cleanup_age backed up the + * horizons further than oldest_considered_running. Fix. + */ + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->shared_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->catalog_oldest_nonremovable); + h->oldest_considered_running = + TransactionIdOlder(h->oldest_considered_running, + h->data_oldest_nonremovable); + + /* + * shared horizons have to be at least as old as the oldest visible in + * current db + */ + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->data_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->shared_oldest_nonremovable, + h->catalog_oldest_nonremovable)); + + /* + * Horizons need to ensure that pg_subtrans access is still possible for + * the relevant backends. + */ + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->shared_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->catalog_oldest_nonremovable)); + Assert(TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->data_oldest_nonremovable)); + Assert(!TransactionIdIsValid(h->slot_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_xmin)); + Assert(!TransactionIdIsValid(h->slot_catalog_xmin) || + TransactionIdPrecedesOrEquals(h->oldest_considered_running, + h->slot_catalog_xmin)); + + /* update approximate horizons with the computed horizons */ + GlobalVisUpdateApply(h); +} + +/* + * Return the oldest XID for which deleted tuples must be preserved in the + * passed table. + * + * If rel is not NULL the horizon may be considerably more recent than + * otherwise (i.e. fewer tuples will be removable). In the NULL case a horizon + * that is correct (but not optimal) for all relations will be returned. + * + * This is used by VACUUM to decide which deleted tuples must be preserved in + * the passed in table. + */ +TransactionId +GetOldestNonRemovableTransactionId(Relation rel) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* select horizon appropriate for relation */ + if (rel == NULL || rel->rd_rel->relisshared) + return horizons.shared_oldest_nonremovable; + else if (RelationIsAccessibleInLogicalDecoding(rel)) + return horizons.catalog_oldest_nonremovable; + else + return horizons.data_oldest_nonremovable; +} + +/* + * Return the oldest transaction id any currently running backend might still + * consider running. This should not be used for visibility / pruning + * determinations (see GetOldestNonRemovableTransactionId()), but for + * decisions like up to where pg_subtrans can be truncated. + */ +TransactionId +GetOldestTransactionIdConsideredRunning(void) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + return horizons.oldest_considered_running; +} + +/* + * Return the visibility horizons for a hot standby feedback message. + */ +void +GetReplicationHorizons(TransactionId *xmin, TransactionId *catalog_xmin) +{ + ComputeXidHorizonsResult horizons; + + ComputeXidHorizons(&horizons); + + /* + * Don't want to use shared_oldest_nonremovable here, as that contains the + * effect of replication slot's catalog_xmin. We want to send a separate + * feedback for the catalog horizon, so the primary can remove data table + * contents more aggressively. + */ + *xmin = horizons.shared_oldest_nonremovable_raw; + *catalog_xmin = horizons.slot_catalog_xmin; } /* @@ -1544,12 +1841,10 @@ GetMaxSnapshotSubxidCount(void) * current transaction (this is the same as MyPgXact->xmin). * RecentXmin: the xmin computed for the most recent snapshot. XIDs * older than this are known not running any more. - * RecentGlobalXmin: the global xmin (oldest TransactionXmin across all - * running transactions, except those running LAZY VACUUM). This is - * the same computation done by - * GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM). - * RecentGlobalDataXmin: the global xmin for non-catalog tables - * >= RecentGlobalXmin + * + * And try to advance the bounds of GlobalVisSharedRels, + * GlobalVisCatalogRels, GlobalVisDataRels for + * the benefit GlobalVis*. * * Note: this function should probably not be called with an argument that's * not statically allocated (see xip allocation below). @@ -1560,12 +1855,12 @@ GetSnapshotData(Snapshot snapshot) ProcArrayStruct *arrayP = procArray; TransactionId xmin; TransactionId xmax; - TransactionId globalxmin; int index; int count = 0; int subcount = 0; bool suboverflowed = false; FullTransactionId latest_completed; + TransactionId oldestxid; TransactionId replication_slot_xmin = InvalidTransactionId; TransactionId replication_slot_catalog_xmin = InvalidTransactionId; @@ -1610,13 +1905,15 @@ GetSnapshotData(Snapshot snapshot) LWLockAcquire(ProcArrayLock, LW_SHARED); latest_completed = ShmemVariableCache->latestCompletedXid; + oldestxid = ShmemVariableCache->oldestXid; + /* xmax is always latestCompletedXid + 1 */ xmax = XidFromFullTransactionId(latest_completed); TransactionIdAdvance(xmax); Assert(TransactionIdIsNormal(xmax)); /* initialize xmin calculation with xmax */ - globalxmin = xmin = xmax; + xmin = xmax; snapshot->takenDuringRecovery = RecoveryInProgress(); @@ -1645,12 +1942,6 @@ GetSnapshotData(Snapshot snapshot) (PROC_IN_LOGICAL_DECODING | PROC_IN_VACUUM)) continue; - /* Update globalxmin to be the smallest valid xmin */ - xid = UINT32_ACCESS_ONCE(pgxact->xmin); - if (TransactionIdIsNormal(xid) && - NormalTransactionIdPrecedes(xid, globalxmin)) - globalxmin = xid; - /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(pgxact->xid); @@ -1766,34 +2057,78 @@ GetSnapshotData(Snapshot snapshot) LWLockRelease(ProcArrayLock); - /* - * Update globalxmin to include actual process xids. This is a slightly - * different way of computing it than GetOldestXmin uses, but should give - * the same result. - */ - if (TransactionIdPrecedes(xmin, globalxmin)) - globalxmin = xmin; + /* maintain state for GlobalVis* */ + { + TransactionId def_vis_xid; + TransactionId def_vis_xid_data; + FullTransactionId def_vis_fxid; + FullTransactionId def_vis_fxid_data; + FullTransactionId oldestfxid; - /* Update global variables too */ - RecentGlobalXmin = globalxmin - vacuum_defer_cleanup_age; - if (!TransactionIdIsNormal(RecentGlobalXmin)) - RecentGlobalXmin = FirstNormalTransactionId; + /* + * Converting oldestXid is only safe when xid horizon cannot advance, + * i.e. holding locks. While we don't hold the lock anymore, all the + * necessary data has been gathered with lock held. + */ + oldestfxid = FullXidRelativeTo(latest_completed, oldestxid); - /* Check whether there's a replication slot requiring an older xmin. */ - if (TransactionIdIsValid(replication_slot_xmin) && - NormalTransactionIdPrecedes(replication_slot_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_xmin; + /* apply vacuum_defer_cleanup_age */ + def_vis_xid_data = + TransactionIdRetreatedBy(xmin, vacuum_defer_cleanup_age); - /* Non-catalog tables can be vacuumed if older than this xid */ - RecentGlobalDataXmin = RecentGlobalXmin; + /* Check whether there's a replication slot requiring an older xmin. */ + def_vis_xid_data = + TransactionIdOlder(def_vis_xid_data, replication_slot_xmin); - /* - * Check whether there's a replication slot requiring an older catalog - * xmin. - */ - if (TransactionIdIsNormal(replication_slot_catalog_xmin) && - NormalTransactionIdPrecedes(replication_slot_catalog_xmin, RecentGlobalXmin)) - RecentGlobalXmin = replication_slot_catalog_xmin; + /* + * Rows in non-shared, non-catalog tables possibly could be vacuumed + * if older than this xid. + */ + def_vis_xid = def_vis_xid_data; + + /* + * Check whether there's a replication slot requiring an older catalog + * xmin. + */ + def_vis_xid = + TransactionIdOlder(replication_slot_catalog_xmin, def_vis_xid); + + def_vis_fxid = FullXidRelativeTo(latest_completed, def_vis_xid); + def_vis_fxid_data = FullXidRelativeTo(latest_completed, def_vis_xid_data); + + /* + * Check if we can increase upper bound. As a previous + * GlobalVisUpdate() might have computed more aggressive values, don't + * overwrite them if so. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(def_vis_fxid_data, + GlobalVisDataRels.definitely_needed); + + /* + * Check if we know that we can initialize or increase the lower + * bound. Currently the only cheap way to do so is to use + * ShmemVariableCache->oldestXid as input. + * + * We should definitely be able to do better. We could e.g. put a + * global lower bound value into ShmemVariableCache. + */ + GlobalVisSharedRels.maybe_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + oldestfxid); + GlobalVisCatalogRels.maybe_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + oldestfxid); + GlobalVisDataRels.maybe_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + oldestfxid); + } RecentXmin = xmin; @@ -3291,6 +3626,255 @@ DisplayXidCache(void) } #endif /* XIDCACHE_DEBUG */ +/* + * If rel != NULL, return test state appropriate for relation, otherwise + * return state usable for all relations. The latter may consider XIDs as + * not-yet-visible-to-everyone that a state for a specific relation would + * already consider visible-to-everyone. + * + * This needs to be called while a snapshot is active or registered, otherwise + * there are wraparound and other dangers. + * + * See comment for GlobalVisState for details. + */ +GlobalVisState * +GlobalVisTestFor(Relation rel) +{ + bool need_shared; + bool need_catalog; + GlobalVisState *state; + + /* XXX: we should assert that a snapshot is pushed or registered */ + Assert(RecentXmin); + + if (!rel) + need_shared = need_catalog = true; + else + { + /* + * Other kinds currently don't contain xids, nor always the necessary + * logical decoding markers. + */ + Assert(rel->rd_rel->relkind == RELKIND_RELATION || + rel->rd_rel->relkind == RELKIND_MATVIEW || + rel->rd_rel->relkind == RELKIND_TOASTVALUE); + + need_shared = rel->rd_rel->relisshared || RecoveryInProgress(); + need_catalog = IsCatalogRelation(rel) || RelationIsAccessibleInLogicalDecoding(rel); + } + + if (need_shared) + state = &GlobalVisSharedRels; + else if (need_catalog) + state = &GlobalVisCatalogRels; + else + state = &GlobalVisDataRels; + + Assert(FullTransactionIdIsValid(state->definitely_needed) && + FullTransactionIdIsValid(state->maybe_needed)); + + return state; +} + +/* + * Return true if it's worth updating the accurate maybe_needed boundary. + * + * As it is somewhat expensive to determine xmin horizons, we don't want to + * repeatedly do so when there is a low likelihood of it being beneficial. + * + * The current heuristic is that we update only if RecentXmin has changed + * since the last update. If the oldest currently running transaction has not + * finished, it is unlikely that recomputing the horizon would be useful. + */ +static bool +GlobalVisTestShouldUpdate(GlobalVisState *state) +{ + /* hasn't been updated yet */ + if (!TransactionIdIsValid(ComputeXidHorizonsResultLastXmin)) + return true; + + /* + * If the maybe_needed/definitely_needed boundaries are the same, it's + * unlikely to be beneficial to refresh boundaries. + */ + if (FullTransactionIdFollowsOrEquals(state->maybe_needed, + state->definitely_needed)) + return false; + + /* does the last snapshot built have a different xmin? */ + return RecentXmin != ComputeXidHorizonsResultLastXmin; +} + +static void +GlobalVisUpdateApply(ComputeXidHorizonsResult *horizons) +{ + GlobalVisSharedRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->shared_oldest_nonremovable); + GlobalVisCatalogRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->catalog_oldest_nonremovable); + GlobalVisDataRels.maybe_needed = + FullXidRelativeTo(horizons->latest_completed, + horizons->data_oldest_nonremovable); + + /* + * In longer running transactions it's possible that transactions we + * previously needed to treat as running aren't around anymore. So update + * definitely_needed to not be earlier than maybe_needed. + */ + GlobalVisSharedRels.definitely_needed = + FullTransactionIdNewer(GlobalVisSharedRels.maybe_needed, + GlobalVisSharedRels.definitely_needed); + GlobalVisCatalogRels.definitely_needed = + FullTransactionIdNewer(GlobalVisCatalogRels.maybe_needed, + GlobalVisCatalogRels.definitely_needed); + GlobalVisDataRels.definitely_needed = + FullTransactionIdNewer(GlobalVisDataRels.maybe_needed, + GlobalVisDataRels.definitely_needed); + + ComputeXidHorizonsResultLastXmin = RecentXmin; +} + +/* + * Update boundaries in GlobalVis{Shared,Catalog, Data}Rels + * using ComputeXidHorizons(). + */ +static void +GlobalVisUpdate(void) +{ + ComputeXidHorizonsResult horizons; + + /* updates the horizons as a side-effect */ + ComputeXidHorizons(&horizons); +} + +/* + * Return true if no snapshot still considers fxid to be running. + * + * The state passed needs to have been initialized for the relation fxid is + * from (NULL is also OK), otherwise the result may not be correct. + * + * See comment for GlobalVisState for details. + */ +bool +GlobalVisTestIsRemovableFullXid(GlobalVisState *state, + FullTransactionId fxid) +{ + /* + * If fxid is older than maybe_needed bound, it definitely is visible to + * everyone. + */ + if (FullTransactionIdPrecedes(fxid, state->maybe_needed)) + return true; + + /* + * If fxid is >= definitely_needed bound, it is very likely to still be + * considered running. + */ + if (FullTransactionIdFollowsOrEquals(fxid, state->definitely_needed)) + return false; + + /* + * fxid is between maybe_needed and definitely_needed, i.e. there might or + * might not exist a snapshot considering fxid running. If it makes sense, + * update boundaries and recheck. + */ + if (GlobalVisTestShouldUpdate(state)) + { + GlobalVisUpdate(); + + Assert(FullTransactionIdPrecedes(fxid, state->definitely_needed)); + + return FullTransactionIdPrecedes(fxid, state->maybe_needed); + } + else + return false; +} + +/* + * Wrapper around GlobalVisTestIsRemovableFullXid() for 32bit xids. + * + * It is crucial that this only gets called for xids from a source that + * protects against xid wraparounds (e.g. from a table and thus protected by + * relfrozenxid). + */ +bool +GlobalVisTestIsRemovableXid(GlobalVisState *state, TransactionId xid) +{ + FullTransactionId fxid; + + /* + * Convert 32 bit argument to FullTransactionId. We can do so safely + * because we know the xid has to, at the very least, be between + * [oldestXid, nextFullXid), i.e. within 2 billion of xid. To avoid taking + * a lock to determine either, we can just compare with + * state->definitely_needed, which was based on those value at the time + * the current snapshot was built. + */ + fxid = FullXidRelativeTo(state->definitely_needed, xid); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Return FullTransactionId below which all transactions are not considered + * running anymore. + * + * Note: This is less efficient than testing with + * GlobalVisTestIsRemovableFullXid as it likely requires building an accurate + * cutoff, even in the case all the XIDs compared with the cutoff are outside + * [maybe_needed, definitely_needed). + */ +FullTransactionId +GlobalVisTestNonRemovableFullHorizon(GlobalVisState *state) +{ + /* acquire accurate horizon if not already done */ + if (GlobalVisTestShouldUpdate(state)) + GlobalVisUpdate(); + + return state->maybe_needed; +} + +/* Convenience wrapper around GlobalVisTestNonRemovableFullHorizon */ +TransactionId +GlobalVisTestNonRemovableHorizon(GlobalVisState *state) +{ + FullTransactionId cutoff; + + cutoff = GlobalVisTestNonRemovableFullHorizon(state); + + return XidFromFullTransactionId(cutoff); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableFullXid(), see their comments. + */ +bool +GlobalVisIsRemovableFullXid(Relation rel, FullTransactionId fxid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableFullXid(state, fxid); +} + +/* + * Convenience wrapper around GlobalVisTestFor() and + * GlobalVisTestIsRemovableXid(), see their comments. + */ +bool +GlobalVisCheckRemovableXid(Relation rel, TransactionId xid) +{ + GlobalVisState *state; + + state = GlobalVisTestFor(rel); + + return GlobalVisTestIsRemovableXid(state, xid); +} + /* * Convert a 32 bit transaction id into 64 bit transaction id, by assuming it * is within MaxTransactionId / 2 of XidFromFullTransactionId(rel). diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index 53d974125fd..00c7afc66fc 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -5786,14 +5786,15 @@ get_actual_variable_endpoint(Relation heapRel, * recent); that case motivates not using SnapshotAny here. * * A crucial point here is that SnapshotNonVacuumable, with - * RecentGlobalXmin as horizon, yields the inverse of the condition that - * the indexscan will use to decide that index entries are killable (see - * heap_hot_search_buffer()). Therefore, if the snapshot rejects a tuple - * (or more precisely, all tuples of a HOT chain) and we have to continue - * scanning past it, we know that the indexscan will mark that index entry - * killed. That means that the next get_actual_variable_endpoint() call - * will not have to re-consider that index entry. In this way we avoid - * repetitive work when this function is used a lot during planning. + * GlobalVisTestFor(heapRel) as horizon, yields the inverse of the + * condition that the indexscan will use to decide that index entries are + * killable (see heap_hot_search_buffer()). Therefore, if the snapshot + * rejects a tuple (or more precisely, all tuples of a HOT chain) and we + * have to continue scanning past it, we know that the indexscan will mark + * that index entry killed. That means that the next + * get_actual_variable_endpoint() call will not have to re-consider that + * index entry. In this way we avoid repetitive work when this function + * is used a lot during planning. * * But using SnapshotNonVacuumable creates a hazard of its own. In a * recently-created index, some index entries may point at "broken" HOT @@ -5805,7 +5806,8 @@ get_actual_variable_endpoint(Relation heapRel, * or could even be NULL. We avoid this hazard because we take the data * from the index entry not the heap. */ - InitNonVacuumableSnapshot(SnapshotNonVacuumable, RecentGlobalXmin); + InitNonVacuumableSnapshot(SnapshotNonVacuumable, + GlobalVisTestFor(heapRel)); index_scan = index_beginscan(heapRel, indexRel, &SnapshotNonVacuumable, diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index f4247ea70d5..893be2f3ddb 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -722,6 +722,10 @@ InitPostgres(const char *in_dbname, Oid dboid, const char *username, * is critical for anything that reads heap pages, because HOT may decide * to prune them even if the process doesn't attempt to modify any * tuples.) + * + * FIXME: This comment is inaccurate / the code buggy. A snapshot that is + * not pushed/active does not reliably prevent HOT pruning (->xmin could + * e.g. be cleared when cache invalidations are processed). */ if (!bootstrap) { diff --git a/src/backend/utils/time/snapmgr.c b/src/backend/utils/time/snapmgr.c index 6b6c8571e23..76578868cf9 100644 --- a/src/backend/utils/time/snapmgr.c +++ b/src/backend/utils/time/snapmgr.c @@ -157,16 +157,9 @@ static Snapshot HistoricSnapshot = NULL; * These are updated by GetSnapshotData. We initialize them this way * for the convenience of TransactionIdIsInProgress: even in bootstrap * mode, we don't want it to say that BootstrapTransactionId is in progress. - * - * RecentGlobalXmin and RecentGlobalDataXmin are initialized to - * InvalidTransactionId, to ensure that no one tries to use a stale - * value. Readers should ensure that it has been set to something else - * before using it. */ TransactionId TransactionXmin = FirstNormalTransactionId; TransactionId RecentXmin = FirstNormalTransactionId; -TransactionId RecentGlobalXmin = InvalidTransactionId; -TransactionId RecentGlobalDataXmin = InvalidTransactionId; /* (table, ctid) => (cmin, cmax) mapping during timetravel */ static HTAB *tuplecid_data = NULL; @@ -581,9 +574,7 @@ SetTransactionSnapshot(Snapshot sourcesnap, VirtualTransactionId *sourcevxid, * Even though we are not going to use the snapshot it computes, we must * call GetSnapshotData, for two reasons: (1) to be sure that * CurrentSnapshotData's XID arrays have been allocated, and (2) to update - * RecentXmin and RecentGlobalXmin. (We could alternatively include those - * two variables in exported snapshot files, but it seems better to have - * snapshot importers compute reasonably up-to-date values for them.) + * the state for GlobalVis*. */ CurrentSnapshot = GetSnapshotData(&CurrentSnapshotData); @@ -956,36 +947,6 @@ xmin_cmp(const pairingheap_node *a, const pairingheap_node *b, void *arg) return 0; } -/* - * Get current RecentGlobalXmin value, as a FullTransactionId. - */ -FullTransactionId -GetFullRecentGlobalXmin(void) -{ - FullTransactionId nextxid_full; - uint32 nextxid_epoch; - TransactionId nextxid_xid; - uint32 epoch; - - Assert(TransactionIdIsNormal(RecentGlobalXmin)); - - /* - * Compute the epoch from the next XID's epoch. This relies on the fact - * that RecentGlobalXmin must be within the 2 billion XID horizon from the - * next XID. - */ - nextxid_full = ReadNextFullTransactionId(); - nextxid_epoch = EpochFromFullTransactionId(nextxid_full); - nextxid_xid = XidFromFullTransactionId(nextxid_full); - - if (RecentGlobalXmin > nextxid_xid) - epoch = nextxid_epoch - 1; - else - epoch = nextxid_epoch; - - return FullTransactionIdFromEpochAndXid(epoch, RecentGlobalXmin); -} - /* * SnapshotResetXmin * @@ -1753,106 +1714,157 @@ GetOldSnapshotThresholdTimestamp(void) return threshold_timestamp; } -static void +void SetOldSnapshotThresholdTimestamp(TimestampTz ts, TransactionId xlimit) { SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + Assert(oldSnapshotControl->threshold_timestamp <= ts); + Assert(TransactionIdPrecedesOrEquals(oldSnapshotControl->threshold_xid, xlimit)); oldSnapshotControl->threshold_timestamp = ts; oldSnapshotControl->threshold_xid = xlimit; SpinLockRelease(&oldSnapshotControl->mutex_threshold); } +/* + * XXX: Magic to keep old_snapshot_threshold tests appear "working". They + * currently are broken, and discussion of what to do about them is + * ongoing. See + * https://www.postgresql.org/message-id/20200403001235.e6jfdll3gh2ygbuc%40alap3.anarazel.de + */ +void +SnapshotTooOldMagicForTest(void) +{ + TimestampTz ts = GetSnapshotCurrentTimestamp(); + + Assert(old_snapshot_threshold == 0); + + ts -= 5 * USECS_PER_SEC; + + SpinLockAcquire(&oldSnapshotControl->mutex_threshold); + oldSnapshotControl->threshold_timestamp = ts; + SpinLockRelease(&oldSnapshotControl->mutex_threshold); +} + +/* + * If there is a valid mapping for the timestamp, set *xlimitp to + * that. Returns whether there is such a mapping. + */ +static bool +GetOldSnapshotFromTimeMapping(TimestampTz ts, TransactionId *xlimitp) +{ + bool in_mapping = false; + + Assert(ts == AlignTimestampToMinuteBoundary(ts)); + + LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); + + if (oldSnapshotControl->count_used > 0 + && ts >= oldSnapshotControl->head_timestamp) + { + int offset; + + offset = ((ts - oldSnapshotControl->head_timestamp) + / USECS_PER_MINUTE); + if (offset > oldSnapshotControl->count_used - 1) + offset = oldSnapshotControl->count_used - 1; + offset = (oldSnapshotControl->head_offset + offset) + % OLD_SNAPSHOT_TIME_MAP_ENTRIES; + + *xlimitp = oldSnapshotControl->xid_by_minute[offset]; + + in_mapping = true; + } + + LWLockRelease(OldSnapshotTimeMapLock); + + return in_mapping; +} + /* * TransactionIdLimitedForOldSnapshots * - * Apply old snapshot limit, if any. This is intended to be called for page - * pruning and table vacuuming, to allow old_snapshot_threshold to override - * the normal global xmin value. Actual testing for snapshot too old will be - * based on whether a snapshot timestamp is prior to the threshold timestamp - * set in this function. + * Apply old snapshot limit. This is intended to be called for page pruning + * and table vacuuming, to allow old_snapshot_threshold to override the normal + * global xmin value. Actual testing for snapshot too old will be based on + * whether a snapshot timestamp is prior to the threshold timestamp set in + * this function. + * + * If the limited horizon allows a cleanup action that otherwise would not be + * possible, SetOldSnapshotThresholdTimestamp(*limit_ts, *limit_xid) needs to + * be called before that cleanup action. */ -TransactionId +bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, - Relation relation) + Relation relation, + TransactionId *limit_xid, + TimestampTz *limit_ts) { - if (TransactionIdIsNormal(recentXmin) - && old_snapshot_threshold >= 0 - && RelationAllowsEarlyPruning(relation)) + TimestampTz ts; + TransactionId xlimit = recentXmin; + TransactionId latest_xmin; + TimestampTz next_map_update_ts; + TransactionId threshold_timestamp; + TransactionId threshold_xid; + + Assert(TransactionIdIsNormal(recentXmin)); + Assert(OldSnapshotThresholdActive()); + Assert(limit_ts != NULL && limit_xid != NULL); + + if (!RelationAllowsEarlyPruning(relation)) + return false; + + ts = GetSnapshotCurrentTimestamp(); + + SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); + latest_xmin = oldSnapshotControl->latest_xmin; + next_map_update_ts = oldSnapshotControl->next_map_update; + SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); + + /* + * Zero threshold always overrides to latest xmin, if valid. Without + * some heuristic it will find its own snapshot too old on, for + * example, a simple UPDATE -- which would make it useless for most + * testing, but there is no principled way to ensure that it doesn't + * fail in this way. Use a five-second delay to try to get useful + * testing behavior, but this may need adjustment. + */ + if (old_snapshot_threshold == 0) { - TimestampTz ts = GetSnapshotCurrentTimestamp(); - TransactionId xlimit = recentXmin; - TransactionId latest_xmin; - TimestampTz update_ts; - bool same_ts_as_threshold = false; - - SpinLockAcquire(&oldSnapshotControl->mutex_latest_xmin); - latest_xmin = oldSnapshotControl->latest_xmin; - update_ts = oldSnapshotControl->next_map_update; - SpinLockRelease(&oldSnapshotControl->mutex_latest_xmin); - - /* - * Zero threshold always overrides to latest xmin, if valid. Without - * some heuristic it will find its own snapshot too old on, for - * example, a simple UPDATE -- which would make it useless for most - * testing, but there is no principled way to ensure that it doesn't - * fail in this way. Use a five-second delay to try to get useful - * testing behavior, but this may need adjustment. - */ - if (old_snapshot_threshold == 0) - { - if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) - && TransactionIdFollows(latest_xmin, xlimit)) - xlimit = latest_xmin; - - ts -= 5 * USECS_PER_SEC; - SetOldSnapshotThresholdTimestamp(ts, xlimit); - - return xlimit; - } + if (TransactionIdPrecedes(latest_xmin, MyPgXact->xmin) + && TransactionIdFollows(latest_xmin, xlimit)) + xlimit = latest_xmin; + ts -= 5 * USECS_PER_SEC; + } + else + { ts = AlignTimestampToMinuteBoundary(ts) - (old_snapshot_threshold * USECS_PER_MINUTE); /* Check for fast exit without LW locking. */ SpinLockAcquire(&oldSnapshotControl->mutex_threshold); - if (ts == oldSnapshotControl->threshold_timestamp) - { - xlimit = oldSnapshotControl->threshold_xid; - same_ts_as_threshold = true; - } + threshold_timestamp = oldSnapshotControl->threshold_timestamp; + threshold_xid = oldSnapshotControl->threshold_xid; SpinLockRelease(&oldSnapshotControl->mutex_threshold); - if (!same_ts_as_threshold) + if (ts == threshold_timestamp) + { + /* + * Current timestamp is in same bucket as the the last limit that + * was applied. Reuse. + */ + xlimit = threshold_xid; + } + else if (ts == next_map_update_ts) + { + /* + * FIXME: This branch is super iffy - but that should probably + * fixed separately. + */ + xlimit = latest_xmin; + } + else if (GetOldSnapshotFromTimeMapping(ts, &xlimit)) { - if (ts == update_ts) - { - xlimit = latest_xmin; - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - else - { - LWLockAcquire(OldSnapshotTimeMapLock, LW_SHARED); - - if (oldSnapshotControl->count_used > 0 - && ts >= oldSnapshotControl->head_timestamp) - { - int offset; - - offset = ((ts - oldSnapshotControl->head_timestamp) - / USECS_PER_MINUTE); - if (offset > oldSnapshotControl->count_used - 1) - offset = oldSnapshotControl->count_used - 1; - offset = (oldSnapshotControl->head_offset + offset) - % OLD_SNAPSHOT_TIME_MAP_ENTRIES; - xlimit = oldSnapshotControl->xid_by_minute[offset]; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - SetOldSnapshotThresholdTimestamp(ts, xlimit); - } - - LWLockRelease(OldSnapshotTimeMapLock); - } } /* @@ -1867,12 +1879,18 @@ TransactionIdLimitedForOldSnapshots(TransactionId recentXmin, if (TransactionIdIsNormal(latest_xmin) && TransactionIdPrecedes(latest_xmin, xlimit)) xlimit = latest_xmin; - - if (NormalTransactionIdFollows(xlimit, recentXmin)) - return xlimit; } - return recentXmin; + if (TransactionIdIsValid(xlimit) && + TransactionIdFollowsOrEquals(xlimit, recentXmin)) + { + *limit_ts = ts; + *limit_xid = xlimit; + + return true; + } + + return false; } /* diff --git a/contrib/amcheck/verify_nbtree.c b/contrib/amcheck/verify_nbtree.c index 635ece73b35..5f3de3c0b7f 100644 --- a/contrib/amcheck/verify_nbtree.c +++ b/contrib/amcheck/verify_nbtree.c @@ -434,10 +434,10 @@ bt_check_every_level(Relation rel, Relation heaprel, bool heapkeyspace, RelationGetRelationName(rel)); /* - * RecentGlobalXmin assertion matches index_getnext_tid(). See note on - * RecentGlobalXmin/B-Tree page deletion. + * This assertion matches the one in index_getnext_tid(). See page + * recycling/"visible to everyone" notes in nbtree README. */ - Assert(TransactionIdIsValid(RecentGlobalXmin)); + Assert(TransactionIdIsValid(RecentXmin)); /* * Initialize state for entire verification operation @@ -1581,7 +1581,7 @@ bt_right_page_check_scankey(BtreeCheckState *state) * does not occur until no possible index scan could land on the page. * Index scans can follow links with nothing more than their snapshot as * an interlock and be sure of at least that much. (See page - * recycling/RecentGlobalXmin notes in nbtree README.) + * recycling/"visible to everyone" notes in nbtree README.) * * Furthermore, it's okay if we follow a rightlink and find a half-dead or * dead (ignorable) page one or more times. There will either be a diff --git a/contrib/pg_visibility/pg_visibility.c b/contrib/pg_visibility/pg_visibility.c index e731161734a..e8cdea7e283 100644 --- a/contrib/pg_visibility/pg_visibility.c +++ b/contrib/pg_visibility/pg_visibility.c @@ -563,17 +563,14 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) BufferAccessStrategy bstrategy = GetAccessStrategy(BAS_BULKREAD); TransactionId OldestXmin = InvalidTransactionId; - if (all_visible) - { - /* Don't pass rel; that will fail in recovery. */ - OldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); - } - rel = relation_open(relid, AccessShareLock); /* Only some relkinds have a visibility map */ check_relation_relkind(rel); + if (all_visible) + OldestXmin = GetOldestNonRemovableTransactionId(rel); + nblocks = RelationGetNumberOfBlocks(rel); /* @@ -679,11 +676,12 @@ collect_corrupt_items(Oid relid, bool all_visible, bool all_frozen) * From a concurrency point of view, it sort of sucks to * retake ProcArrayLock here while we're holding the buffer * exclusively locked, but it should be safe against - * deadlocks, because surely GetOldestXmin() should never take - * a buffer lock. And this shouldn't happen often, so it's - * worth being careful so as to avoid false positives. + * deadlocks, because surely GetOldestNonRemovableTransactionId() + * should never take a buffer lock. And this shouldn't happen + * often, so it's worth being careful so as to avoid false + * positives. */ - RecomputedOldestXmin = GetOldestXmin(NULL, PROCARRAY_FLAGS_VACUUM); + RecomputedOldestXmin = GetOldestNonRemovableTransactionId(rel); if (!TransactionIdPrecedes(OldestXmin, RecomputedOldestXmin)) record_corrupt_item(items, &tuple.t_self); diff --git a/contrib/pgstattuple/pgstatapprox.c b/contrib/pgstattuple/pgstatapprox.c index dbc0fa11f61..3a99333d443 100644 --- a/contrib/pgstattuple/pgstatapprox.c +++ b/contrib/pgstattuple/pgstatapprox.c @@ -71,7 +71,7 @@ statapprox_heap(Relation rel, output_type *stat) BufferAccessStrategy bstrategy; TransactionId OldestXmin; - OldestXmin = GetOldestXmin(rel, PROCARRAY_FLAGS_VACUUM); + OldestXmin = GetOldestNonRemovableTransactionId(rel); bstrategy = GetAccessStrategy(BAS_BULKREAD); nblocks = RelationGetNumberOfBlocks(rel); diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 7eaaad1e140..b4948ac675f 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -395,6 +395,7 @@ CompositeTypeStmt CompoundAffixFlag CompressionAlgorithm CompressorState +ComputeXidHorizonsResult ConditionVariable ConditionalStack ConfigData @@ -930,6 +931,7 @@ GistSplitVector GistTsVectorOptions GistVacState GlobalTransaction +GlobalVisState GrantRoleStmt GrantStmt GrantTargetType -- 2.25.0.114.g5b0ca878e0 --4il4azyv6rmablac Content-Type: text/x-diff; charset=us-ascii Content-Disposition: attachment; filename="v13-0002-snapshot-scalability-Move-PGXACT-xmin-back-to-PG.patch" ^ permalink raw reply [nested|flat] 29+ messages in thread
* Re: Add LZ4 compression in pg_dump @ 2023-01-14 21:43 Justin Pryzby <[email protected]> 0 siblings, 2 replies; 29+ messages in thread From: Justin Pryzby @ 2023-01-14 21:43 UTC (permalink / raw) To: [email protected]; +Cc: Michael Paquier <[email protected]>; [email protected]; Rachel Heaton <[email protected]> On Sun, Jan 08, 2023 at 01:45:25PM -0600, Justin Pryzby wrote: > On Thu, Dec 22, 2022 at 11:08:59AM -0600, Justin Pryzby wrote: > > There's a couple of lz4 bits which shouldn't be present in 002: file > > extension and comments. > BTW I noticed that cfdopen() was accidentally committed to compress_io.h > in master without being defined anywhere. This was resolved in 69fb29d1a (so now needs to be re-added for this patch series). > pg_compress_specification is being passed by value, but I think it > should be passed as a pointer, as is done everywhere else. ISTM that was an issue with 5e73a6048, affecting a few public and private functions. I wrote a pre-preparatory patch which changes to pass by reference. And addressed a handful of other issues I reported as separate fixup commits. And changed to use LZ4 by default for CI. I also rebased my 2 year old patch to support zstd in pg_dump. I hope it can finally added for v16. I'll send it for the next CF if these patches progress. One more thing: some comments still refer to the cfopen API, which this patch removes. > There were "LZ4" comments and file extension stuff in the preparatory > commit. But now it seems like you *removed* them in the LZ4 commit > (where it actually belongs) rather than *moving* it from the > prior/parent commit *to* the lz4 commit. I recommend to run something > like "git diff @{1}" whenever doing this kind of patch surgery. TODO > Maybe other places that check if (compression==PG_COMPRESSION_GZIP) > should maybe change to say compression!=NONE? > > _PrepParallelRestore() references ".gz", so I think it needs to be > retrofitted to handle .lz4. Ideally, that's built into a struct or list > of file extensions to try. Maybe compression.h should have a function > to return the file extension of a given algorithm. I'm planning to send > a patch for zstd, and hoping its changes will be minimized by these > preparatory commits. TODO > I think it's confusing to have two functions, one named > InitCompressLZ4() and InitCompressorLZ4(). TODO > pg_compress_algorithm is being writen directly into the pg_dump header. > Currently, I think that's not an externally-visible value (it could be > renumbered, theoretically even in a minor release). Maybe there should > be a "private" enum for encoding the pg_dump header, similar to > WAL_COMPRESSION_LZ4 vs BKPIMAGE_COMPRESS_LZ4 ? Or else a comment there > should warn that the values are encoded in pg_dump, and must never be > changed. Michael, WDYT ? -- Justin Attachments: [text/x-diff] 0001-pg_dump-pass-pg_compress_specification-as-a-pointer.patch (11.9K, ../../[email protected]/2-0001-pg_dump-pass-pg_compress_specification-as-a-pointer.patch) download | inline diff: From 3105d480ab82093ca2873e423782f5b2edd9fbb7 Mon Sep 17 00:00:00 2001 From: Justin Pryzby <[email protected]> Date: Sat, 14 Jan 2023 10:58:23 -0600 Subject: [PATCH 1/7] pg_dump: pass pg_compress_specification as a pointer.. ..as is done everywhere else. --- src/bin/pg_dump/compress_io.c | 30 +++++++++++++-------------- src/bin/pg_dump/compress_io.h | 8 +++---- src/bin/pg_dump/pg_backup.h | 2 +- src/bin/pg_dump/pg_backup_archiver.c | 12 +++++------ src/bin/pg_dump/pg_backup_custom.c | 6 +++--- src/bin/pg_dump/pg_backup_directory.c | 8 +++---- src/bin/pg_dump/pg_dump.c | 2 +- 7 files changed, 34 insertions(+), 34 deletions(-) diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index 7a2c80bbc4c..62a9527fa48 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -94,19 +94,19 @@ static void WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, /* Allocate a new compressor */ CompressorState * -AllocateCompressor(const pg_compress_specification compression_spec, +AllocateCompressor(const pg_compress_specification *compression_spec, WriteFunc writeF) { CompressorState *cs; #ifndef HAVE_LIBZ - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) + if (compression_spec->algorithm == PG_COMPRESSION_GZIP) pg_fatal("this build does not support compression with %s", "gzip"); #endif cs = (CompressorState *) pg_malloc0(sizeof(CompressorState)); cs->writeF = writeF; - cs->compression_spec = compression_spec; + cs->compression_spec = *compression_spec; // XXX /* * Perform compression algorithm specific initialization. @@ -125,12 +125,12 @@ AllocateCompressor(const pg_compress_specification compression_spec, */ void ReadDataFromArchive(ArchiveHandle *AH, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, ReadFunc readF) { - if (compression_spec.algorithm == PG_COMPRESSION_NONE) + if (compression_spec->algorithm == PG_COMPRESSION_NONE) ReadDataFromArchiveNone(AH, readF); - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) + if (compression_spec->algorithm == PG_COMPRESSION_GZIP) { #ifdef HAVE_LIBZ ReadDataFromArchiveZlib(AH, readF); @@ -432,13 +432,13 @@ cfopen_read(const char *path, const char *mode) if (hasSuffix(path, ".gz")) { compression_spec.algorithm = PG_COMPRESSION_GZIP; - fp = cfopen(path, mode, compression_spec); + fp = cfopen(path, mode, &compression_spec); } else #endif { compression_spec.algorithm = PG_COMPRESSION_NONE; - fp = cfopen(path, mode, compression_spec); + fp = cfopen(path, mode, &compression_spec); #ifdef HAVE_LIBZ if (fp == NULL) { @@ -446,7 +446,7 @@ cfopen_read(const char *path, const char *mode) fname = psprintf("%s.gz", path); compression_spec.algorithm = PG_COMPRESSION_GZIP; - fp = cfopen(fname, mode, compression_spec); + fp = cfopen(fname, mode, &compression_spec); free_keep_errno(fname); } #endif @@ -467,11 +467,11 @@ cfopen_read(const char *path, const char *mode) */ cfp * cfopen_write(const char *path, const char *mode, - const pg_compress_specification compression_spec) + const pg_compress_specification *compression_spec) { cfp *fp; - if (compression_spec.algorithm == PG_COMPRESSION_NONE) + if (compression_spec->algorithm == PG_COMPRESSION_NONE) fp = cfopen(path, mode, compression_spec); else { @@ -497,20 +497,20 @@ cfopen_write(const char *path, const char *mode, */ cfp * cfopen(const char *path, const char *mode, - const pg_compress_specification compression_spec) + const pg_compress_specification *compression_spec) { cfp *fp = pg_malloc(sizeof(cfp)); - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) + if (compression_spec->algorithm == PG_COMPRESSION_GZIP) { #ifdef HAVE_LIBZ - if (compression_spec.level != Z_DEFAULT_COMPRESSION) + if (compression_spec->level != Z_DEFAULT_COMPRESSION) { /* user has specified a compression level, so tell zlib to use it */ char mode_compression[32]; snprintf(mode_compression, sizeof(mode_compression), "%s%d", - mode, compression_spec.level); + mode, compression_spec->level); fp->compressedfp = gzopen(path, mode_compression); } else diff --git a/src/bin/pg_dump/compress_io.h b/src/bin/pg_dump/compress_io.h index a429dc4789d..34f4e5e1e14 100644 --- a/src/bin/pg_dump/compress_io.h +++ b/src/bin/pg_dump/compress_io.h @@ -40,10 +40,10 @@ typedef size_t (*ReadFunc) (ArchiveHandle *AH, char **buf, size_t *buflen); /* struct definition appears in compress_io.c */ typedef struct CompressorState CompressorState; -extern CompressorState *AllocateCompressor(const pg_compress_specification compression_spec, +extern CompressorState *AllocateCompressor(const pg_compress_specification *compression_spec, WriteFunc writeF); extern void ReadDataFromArchive(ArchiveHandle *AH, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, ReadFunc readF); extern void WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, const void *data, size_t dLen); @@ -53,10 +53,10 @@ extern void EndCompressor(ArchiveHandle *AH, CompressorState *cs); typedef struct cfp cfp; extern cfp *cfopen(const char *path, const char *mode, - const pg_compress_specification compression_spec); + const pg_compress_specification *compression_spec); extern cfp *cfopen_read(const char *path, const char *mode); extern cfp *cfopen_write(const char *path, const char *mode, - const pg_compress_specification compression_spec); + const pg_compress_specification *compression_spec); extern int cfread(void *ptr, int size, cfp *fp); extern int cfwrite(const void *ptr, int size, cfp *fp); extern int cfgetc(cfp *fp); diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h index aba780ef4b1..216e24e7ec5 100644 --- a/src/bin/pg_dump/pg_backup.h +++ b/src/bin/pg_dump/pg_backup.h @@ -305,7 +305,7 @@ extern Archive *OpenArchive(const char *FileSpec, const ArchiveFormat fmt); /* Create a new archive */ extern Archive *CreateArchive(const char *FileSpec, const ArchiveFormat fmt, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, bool dosync, ArchiveMode mode, SetupWorkerPtrType setupDumpWorker); diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index 7f7a0f1ce7b..0d91b75c748 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -70,7 +70,7 @@ typedef struct _parallelReadyList static ArchiveHandle *_allocAH(const char *FileSpec, const ArchiveFormat fmt, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, bool dosync, ArchiveMode mode, SetupWorkerPtrType setupWorkerPtr); static void _getObjectDescription(PQExpBuffer buf, const TocEntry *te); @@ -241,7 +241,7 @@ setupRestoreWorker(Archive *AHX) /* Public */ Archive * CreateArchive(const char *FileSpec, const ArchiveFormat fmt, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, bool dosync, ArchiveMode mode, SetupWorkerPtrType setupDumpWorker) @@ -261,7 +261,7 @@ OpenArchive(const char *FileSpec, const ArchiveFormat fmt) pg_compress_specification compression_spec = {0}; compression_spec.algorithm = PG_COMPRESSION_NONE; - AH = _allocAH(FileSpec, fmt, compression_spec, true, + AH = _allocAH(FileSpec, fmt, &compression_spec, true, archModeRead, setupRestoreWorker); return (Archive *) AH; @@ -2214,7 +2214,7 @@ _discoverArchiveFormat(ArchiveHandle *AH) */ static ArchiveHandle * _allocAH(const char *FileSpec, const ArchiveFormat fmt, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, bool dosync, ArchiveMode mode, SetupWorkerPtrType setupWorkerPtr) { @@ -2266,7 +2266,7 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, AH->toc->prev = AH->toc; AH->mode = mode; - AH->compression_spec = compression_spec; + AH->compression_spec = *compression_spec; // XXX AH->dosync = dosync; memset(&(AH->sqlparse), 0, sizeof(AH->sqlparse)); @@ -2281,7 +2281,7 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, * Force stdin/stdout into binary mode if that is what we are using. */ #ifdef WIN32 - if ((fmt != archNull || compression_spec.algorithm != PG_COMPRESSION_NONE) && + if ((fmt != archNull || compression_spec->algorithm != PG_COMPRESSION_NONE) && (AH->fSpec == NULL || strcmp(AH->fSpec, "") == 0)) { if (mode == archModeWrite) diff --git a/src/bin/pg_dump/pg_backup_custom.c b/src/bin/pg_dump/pg_backup_custom.c index d1e54644a94..0e87444de85 100644 --- a/src/bin/pg_dump/pg_backup_custom.c +++ b/src/bin/pg_dump/pg_backup_custom.c @@ -298,7 +298,7 @@ _StartData(ArchiveHandle *AH, TocEntry *te) _WriteByte(AH, BLK_DATA); /* Block type */ WriteInt(AH, te->dumpId); /* For sanity check */ - ctx->cs = AllocateCompressor(AH->compression_spec, _CustomWriteFunc); + ctx->cs = AllocateCompressor(&AH->compression_spec, _CustomWriteFunc); } /* @@ -377,7 +377,7 @@ _StartLO(ArchiveHandle *AH, TocEntry *te, Oid oid) WriteInt(AH, oid); - ctx->cs = AllocateCompressor(AH->compression_spec, _CustomWriteFunc); + ctx->cs = AllocateCompressor(&AH->compression_spec, _CustomWriteFunc); } /* @@ -566,7 +566,7 @@ _PrintTocData(ArchiveHandle *AH, TocEntry *te) static void _PrintData(ArchiveHandle *AH) { - ReadDataFromArchive(AH, AH->compression_spec, _CustomReadFunc); + ReadDataFromArchive(AH, &AH->compression_spec, _CustomReadFunc); } static void diff --git a/src/bin/pg_dump/pg_backup_directory.c b/src/bin/pg_dump/pg_backup_directory.c index 6800c3cceef..ffb8a0e4dd7 100644 --- a/src/bin/pg_dump/pg_backup_directory.c +++ b/src/bin/pg_dump/pg_backup_directory.c @@ -328,7 +328,7 @@ _StartData(ArchiveHandle *AH, TocEntry *te) setFilePath(AH, fname, tctx->filename); ctx->dataFH = cfopen_write(fname, PG_BINARY_W, - AH->compression_spec); + &AH->compression_spec); if (ctx->dataFH == NULL) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -584,7 +584,7 @@ _CloseArchive(ArchiveHandle *AH) /* The TOC is always created uncompressed */ compression_spec.algorithm = PG_COMPRESSION_NONE; - tocFH = cfopen_write(fname, PG_BINARY_W, compression_spec); + tocFH = cfopen_write(fname, PG_BINARY_W, &compression_spec); if (tocFH == NULL) pg_fatal("could not open output file \"%s\": %m", fname); ctx->dataFH = tocFH; @@ -649,7 +649,7 @@ _StartLOs(ArchiveHandle *AH, TocEntry *te) /* The LO TOC file is never compressed */ compression_spec.algorithm = PG_COMPRESSION_NONE; - ctx->LOsTocFH = cfopen_write(fname, "ab", compression_spec); + ctx->LOsTocFH = cfopen_write(fname, "ab", &compression_spec); if (ctx->LOsTocFH == NULL) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -667,7 +667,7 @@ _StartLO(ArchiveHandle *AH, TocEntry *te, Oid oid) snprintf(fname, MAXPGPATH, "%s/blob_%u.dat", ctx->directory, oid); - ctx->dataFH = cfopen_write(fname, PG_BINARY_W, AH->compression_spec); + ctx->dataFH = cfopen_write(fname, PG_BINARY_W, &AH->compression_spec); if (ctx->dataFH == NULL) pg_fatal("could not open output file \"%s\": %m", fname); diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c index 2c0a9699729..20f73729fac 100644 --- a/src/bin/pg_dump/pg_dump.c +++ b/src/bin/pg_dump/pg_dump.c @@ -751,7 +751,7 @@ main(int argc, char **argv) pg_fatal("parallel backup only supported by the directory format"); /* Open the output file */ - fout = CreateArchive(filename, archiveFormat, compression_spec, + fout = CreateArchive(filename, archiveFormat, &compression_spec, dosync, archiveMode, setupDumpWorker); /* Make dump options accessible right away */ -- 2.25.1 [text/x-diff] 0002-Prepare-pg_dump-internals-for-additional-compression.patch (23.5K, ../../[email protected]/3-0002-Prepare-pg_dump-internals-for-additional-compression.patch) download | inline diff: From 6a8f2cd926be4f0b83f2d2d5170cf02a2a825036 Mon Sep 17 00:00:00 2001 From: Georgios Kokolatos <[email protected]> Date: Wed, 21 Dec 2022 09:49:23 +0000 Subject: [PATCH 2/7] Prepare pg_dump internals for additional compression methods. Commit bf9aa490db introduced cfp in compress_io.{c,h} with the intent of unifying compression related code and allow for the introduction of additional archive formats. However, pg_backup_archiver.c was not using that API. This commit teaches pg_backup_archiver.c about it and is using it throughout. --- src/bin/pg_dump/compress_io.c | 391 +++++++++++++++++++-------- src/bin/pg_dump/compress_io.h | 2 + src/bin/pg_dump/pg_backup_archiver.c | 136 ++++------ src/bin/pg_dump/pg_backup_archiver.h | 27 +- 4 files changed, 326 insertions(+), 230 deletions(-) diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index 62a9527fa48..97b18337578 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -56,6 +56,10 @@ #include "compress_io.h" #include "pg_backup_utils.h" +#ifdef HAVE_LIBZ +#include <zlib.h> +#endif + /*---------------------- * Compressor API *---------------------- @@ -128,15 +132,24 @@ ReadDataFromArchive(ArchiveHandle *AH, const pg_compress_specification *compression_spec, ReadFunc readF) { - if (compression_spec->algorithm == PG_COMPRESSION_NONE) - ReadDataFromArchiveNone(AH, readF); - if (compression_spec->algorithm == PG_COMPRESSION_GZIP) + switch (compression_spec->algorithm) { + case PG_COMPRESSION_NONE: + ReadDataFromArchiveNone(AH, readF); + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - ReadDataFromArchiveZlib(AH, readF); + ReadDataFromArchiveZlib(AH, readF); #else - pg_fatal("this build does not support compression with %s", "gzip"); + pg_fatal("this build does not support compression with %s", "gzip"); #endif + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } } @@ -149,6 +162,9 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, { switch (cs->compression_spec.algorithm) { + case PG_COMPRESSION_NONE: + WriteDataToArchiveNone(AH, cs, data, dLen); + break; case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ WriteDataToArchiveZlib(AH, cs, data, dLen); @@ -156,13 +172,11 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, pg_fatal("this build does not support compression with %s", "gzip"); #endif break; - case PG_COMPRESSION_NONE: - WriteDataToArchiveNone(AH, cs, data, dLen); - break; case PG_COMPRESSION_LZ4: - /* fallthrough */ + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; case PG_COMPRESSION_ZSTD: - pg_fatal("invalid compression method"); + pg_fatal("compression with %s is not yet supported", "ZSTD"); break; } } @@ -173,10 +187,26 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, void EndCompressor(ArchiveHandle *AH, CompressorState *cs) { + switch (cs->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (cs->compression_spec.algorithm == PG_COMPRESSION_GZIP) - EndCompressorZlib(AH, cs); + EndCompressorZlib(AH, cs); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + free(cs); } @@ -391,10 +421,8 @@ WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, */ struct cfp { - FILE *uncompressedfp; -#ifdef HAVE_LIBZ - gzFile compressedfp; -#endif + pg_compress_specification compression_spec; + void *fp; }; #ifdef HAVE_LIBZ @@ -490,127 +518,204 @@ cfopen_write(const char *path, const char *mode, } /* - * Opens file 'path' in 'mode'. If compression is GZIP, the file - * is opened with libz gzopen(), otherwise with plain fopen(). + * This is the workhorse for cfopen() or cfdopen(). It opens file 'path' or + * associates a stream 'fd', if 'fd' is a valid descriptor, in 'mode'. The + * descriptor is not dup'ed and it is the caller's responsibility to do so. + * The caller must verify that the 'compress_algorithm' is supported by the + * current build. * * On failure, return NULL with an error code in errno. */ -cfp * -cfopen(const char *path, const char *mode, - const pg_compress_specification *compression_spec) +static cfp * +cfopen_internal(const char *path, int fd, const char *mode, + const pg_compress_specification *compression_spec) { cfp *fp = pg_malloc(sizeof(cfp)); - if (compression_spec->algorithm == PG_COMPRESSION_GZIP) + fp->compression_spec = *compression_spec; + + switch (compression_spec->algorithm) { -#ifdef HAVE_LIBZ - if (compression_spec->level != Z_DEFAULT_COMPRESSION) - { - /* user has specified a compression level, so tell zlib to use it */ - char mode_compression[32]; + case PG_COMPRESSION_NONE: + if (fd >= 0) + fp->fp = fdopen(fd, mode); + else + fp->fp = fopen(path, mode); + if (fp->fp == NULL) + { + free_keep_errno(fp); + fp = NULL; + } - snprintf(mode_compression, sizeof(mode_compression), "%s%d", - mode, compression_spec->level); - fp->compressedfp = gzopen(path, mode_compression); - } - else - { - /* don't specify a level, just use the zlib default */ - fp->compressedfp = gzopen(path, mode); - } + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + if (compression_spec->level != Z_DEFAULT_COMPRESSION) + { + /* + * user has specified a compression level, so tell zlib to use + * it + */ + char mode_compression[32]; + + snprintf(mode_compression, sizeof(mode_compression), "%s%d", + mode, compression_spec->level); + if (fd >= 0) + fp->fp = gzdopen(fd, mode_compression); + // fp->compressedfp = gzopen(path, mode_compression); + else + fp->fp = gzopen(path, mode_compression); + } + else + { + /* don't specify a level, just use the zlib default */ + if (fd >= 0) + fp->fp = gzdopen(fd, mode); + else + fp->fp = gzopen(path, mode); + // fp->compressedfp = gzopen(path, mode); + } - fp->uncompressedfp = NULL; - if (fp->compressedfp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } + if (fp->fp == NULL) + { + free_keep_errno(fp); + fp = NULL; + } #else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - } - else - { -#ifdef HAVE_LIBZ - fp->compressedfp = NULL; + pg_fatal("this build does not support compression with %s", "gzip"); #endif - fp->uncompressedfp = fopen(path, mode); - if (fp->uncompressedfp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } return fp; } +cfp * +cfopen(const char *path, const char *mode, + const pg_compress_specification *compression_spec) +{ + return cfopen_internal(path, -1, mode, compression_spec); +} + +cfp * +cfdopen(int fd, const char *mode, + const pg_compress_specification *compression_spec) +{ + return cfopen_internal(NULL, fd, mode, compression_spec); +} int cfread(void *ptr, int size, cfp *fp) { - int ret; + int ret = 0; if (size == 0) return 0; -#ifdef HAVE_LIBZ - if (fp->compressedfp) + switch (fp->compression_spec.algorithm) { - ret = gzread(fp->compressedfp, ptr, size); - if (ret != size && !gzeof(fp->compressedfp)) - { - int errnum; - const char *errmsg = gzerror(fp->compressedfp, &errnum); + case PG_COMPRESSION_NONE: + ret = fread(ptr, 1, size, (FILE *) fp->fp); + if (ret != size && !feof((FILE *) fp->fp)) + READ_ERROR_EXIT((FILE *) fp->fp); - pg_fatal("could not read from input file: %s", - errnum == Z_ERRNO ? strerror(errno) : errmsg); - } - } - else + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + ret = gzread((gzFile) fp->fp, ptr, size); + if (ret != size && !gzeof((gzFile) fp->fp)) + { + int errnum; + const char *errmsg = gzerror((gzFile) fp->fp, &errnum); + + pg_fatal("could not read from input file: %s", + errnum == Z_ERRNO ? strerror(errno) : errmsg); + } +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - { - ret = fread(ptr, 1, size, fp->uncompressedfp); - if (ret != size && !feof(fp->uncompressedfp)) - READ_ERROR_EXIT(fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } + return ret; } int cfwrite(const void *ptr, int size, cfp *fp) { + int ret = 0; + + switch (fp->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + ret = fwrite(ptr, 1, size, (FILE *) fp->fp); + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (fp->compressedfp) - return gzwrite(fp->compressedfp, ptr, size); - else + ret = gzwrite((gzFile) fp->fp, ptr, size); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - return fwrite(ptr, 1, size, fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + + return ret; } int cfgetc(cfp *fp) { - int ret; + int ret = 0; -#ifdef HAVE_LIBZ - if (fp->compressedfp) + switch (fp->compression_spec.algorithm) { - ret = gzgetc(fp->compressedfp); - if (ret == EOF) - { - if (!gzeof(fp->compressedfp)) - pg_fatal("could not read from input file: %s", strerror(errno)); - else - pg_fatal("could not read from input file: end of file"); - } - } - else + case PG_COMPRESSION_NONE: + ret = fgetc((FILE *) fp->fp); + if (ret == EOF) + READ_ERROR_EXIT((FILE *) fp->fp); + + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + ret = gzgetc((gzFile) fp->fp); + if (ret == EOF) + { + if (!gzeof((gzFile) fp->fp)) + pg_fatal("could not read from input file: %s", strerror(errno)); + else + pg_fatal("could not read from input file: end of file"); + } +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - { - ret = fgetc(fp->uncompressedfp); - if (ret == EOF) - READ_ERROR_EXIT(fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } return ret; @@ -619,65 +724,119 @@ cfgetc(cfp *fp) char * cfgets(cfp *fp, char *buf, int len) { + char *ret = NULL; + + switch (fp->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + ret = fgets(buf, len, (FILE *) fp->fp); + + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (fp->compressedfp) - return gzgets(fp->compressedfp, buf, len); - else + ret = gzgets((gzFile) fp->fp, buf, len); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - return fgets(buf, len, fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + + return ret; } int cfclose(cfp *fp) { - int result; + int ret = 0; if (fp == NULL) { errno = EBADF; return EOF; } -#ifdef HAVE_LIBZ - if (fp->compressedfp) + + switch (fp->compression_spec.algorithm) { - result = gzclose(fp->compressedfp); - fp->compressedfp = NULL; - } - else + case PG_COMPRESSION_NONE: + ret = fclose((FILE *) fp->fp); + fp->fp = NULL; + + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + ret = gzclose((gzFile) fp->fp); + fp->fp = NULL; +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - { - result = fclose(fp->uncompressedfp); - fp->uncompressedfp = NULL; + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } + free_keep_errno(fp); - return result; + return ret; } int cfeof(cfp *fp) { + int ret = 0; + + switch (fp->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + ret = feof((FILE *) fp->fp); + + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (fp->compressedfp) - return gzeof(fp->compressedfp); - else + ret = gzeof((gzFile) fp->fp); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - return feof(fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + + return ret; } const char * get_cfp_error(cfp *fp) { -#ifdef HAVE_LIBZ - if (fp->compressedfp) + if (fp->compression_spec.algorithm == PG_COMPRESSION_GZIP) { +#ifdef HAVE_LIBZ int errnum; - const char *errmsg = gzerror(fp->compressedfp, &errnum); + const char *errmsg = gzerror((gzFile) fp->fp, &errnum); if (errnum != Z_ERRNO) return errmsg; - } +#else + pg_fatal("this build does not support compression with %s", "gzip"); #endif + } + return strerror(errno); } diff --git a/src/bin/pg_dump/compress_io.h b/src/bin/pg_dump/compress_io.h index 34f4e5e1e14..768096c820d 100644 --- a/src/bin/pg_dump/compress_io.h +++ b/src/bin/pg_dump/compress_io.h @@ -54,6 +54,8 @@ typedef struct cfp cfp; extern cfp *cfopen(const char *path, const char *mode, const pg_compress_specification *compression_spec); +extern cfp *cfdopen(int fd, const char *mode, + const pg_compress_specification *compression_spec); extern cfp *cfopen_read(const char *path, const char *mode); extern cfp *cfopen_write(const char *path, const char *mode, const pg_compress_specification *compression_spec); diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index 0d91b75c748..cbe110c917a 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -31,6 +31,7 @@ #endif #include "common/string.h" +#include "compress_io.h" #include "dumputils.h" #include "fe_utils/string_utils.h" #include "lib/stringinfo.h" @@ -43,13 +44,6 @@ #define TEXT_DUMP_HEADER "--\n-- PostgreSQL database dump\n--\n\n" #define TEXT_DUMPALL_HEADER "--\n-- PostgreSQL database cluster dump\n--\n\n" -/* state needed to save/restore an archive's output target */ -typedef struct _outputContext -{ - void *OF; - int gzOut; -} OutputContext; - /* * State for tracking TocEntrys that are ready to process during a parallel * restore. (This used to be a list, and we still call it that, though now @@ -100,9 +94,9 @@ static int RestoringToDB(ArchiveHandle *AH); static void dump_lo_buf(ArchiveHandle *AH); static void dumpTimestamp(ArchiveHandle *AH, const char *msg, time_t tim); static void SetOutput(ArchiveHandle *AH, const char *filename, - const pg_compress_specification compression_spec); -static OutputContext SaveOutput(ArchiveHandle *AH); -static void RestoreOutput(ArchiveHandle *AH, OutputContext savedContext); + const pg_compress_specification *compression_spec); +static cfp *SaveOutput(ArchiveHandle *AH); +static void RestoreOutput(ArchiveHandle *AH, cfp *savedOutput); static int restore_toc_entry(ArchiveHandle *AH, TocEntry *te, bool is_parallel); static void restore_toc_entries_prefork(ArchiveHandle *AH, @@ -277,11 +271,8 @@ CloseArchive(Archive *AHX) AH->ClosePtr(AH); /* Close the output */ - errno = 0; /* in case gzclose() doesn't set it */ - if (AH->gzOut) - res = GZCLOSE(AH->OF); - else if (AH->OF != stdout) - res = fclose(AH->OF); + errno = 0; + res = cfclose(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); @@ -362,8 +353,9 @@ RestoreArchive(Archive *AHX) ArchiveHandle *AH = (ArchiveHandle *) AHX; RestoreOptions *ropt = AH->public.ropt; bool parallel_mode; + bool supports_compression; TocEntry *te; - OutputContext sav; + cfp *sav; AH->stage = STAGE_INITIALIZING; @@ -391,17 +383,24 @@ RestoreArchive(Archive *AHX) /* * Make sure we won't need (de)compression we haven't got */ -#ifndef HAVE_LIBZ - if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP && + supports_compression = true; + if (AH->compression_spec.algorithm != PG_COMPRESSION_NONE && + AH->compression_spec.algorithm == PG_COMPRESSION_GZIP && AH->PrintTocDataPtr != NULL) { for (te = AH->toc->next; te != AH->toc; te = te->next) { if (te->hadDumper && (te->reqs & REQ_DATA) != 0) - pg_fatal("cannot restore from compressed archive (compression not supported in this installation)"); + { +#ifndef HAVE_LIBZ + if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) + supports_compression = false; +#endif + if (supports_compression == false) + pg_fatal("cannot restore from compressed archive (compression not supported in this installation)"); + } } } -#endif /* * Prepare index arrays, so we can assume we have them throughout restore. @@ -469,7 +468,7 @@ RestoreArchive(Archive *AHX) */ sav = SaveOutput(AH); if (ropt->filename || ropt->compression_spec.algorithm != PG_COMPRESSION_NONE) - SetOutput(AH, ropt->filename, ropt->compression_spec); + SetOutput(AH, ropt->filename, &ropt->compression_spec); ahprintf(AH, "--\n-- PostgreSQL database dump\n--\n\n"); @@ -1128,7 +1127,7 @@ PrintTOCSummary(Archive *AHX) TocEntry *te; pg_compress_specification out_compression_spec = {0}; teSection curSection; - OutputContext sav; + cfp *sav; const char *fmtName; char stamp_str[64]; @@ -1137,7 +1136,7 @@ PrintTOCSummary(Archive *AHX) sav = SaveOutput(AH); if (ropt->filename) - SetOutput(AH, ropt->filename, out_compression_spec); + SetOutput(AH, ropt->filename, &out_compression_spec); if (strftime(stamp_str, sizeof(stamp_str), PGDUMP_STRFTIME_FMT, localtime(&AH->createDate)) == 0) @@ -1501,60 +1500,34 @@ archprintf(Archive *AH, const char *fmt,...) static void SetOutput(ArchiveHandle *AH, const char *filename, - const pg_compress_specification compression_spec) + const pg_compress_specification *compression_spec) { - int fn; + const char *mode; + int fn = -1; if (filename) { if (strcmp(filename, "-") == 0) fn = fileno(stdout); - else - fn = -1; } else if (AH->FH) fn = fileno(AH->FH); else if (AH->fSpec) { - fn = -1; filename = AH->fSpec; } else fn = fileno(stdout); - /* If compression explicitly requested, use gzopen */ -#ifdef HAVE_LIBZ - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) - { - char fmode[14]; + if (AH->mode == archModeAppend) + mode = PG_BINARY_A; + else + mode = PG_BINARY_W; - /* Don't use PG_BINARY_x since this is zlib */ - sprintf(fmode, "wb%d", compression_spec.level); - if (fn >= 0) - AH->OF = gzdopen(dup(fn), fmode); - else - AH->OF = gzopen(filename, fmode); - AH->gzOut = 1; - } + if (fn >= 0) + AH->OF = cfdopen(dup(fn), mode, compression_spec); else -#endif - { /* Use fopen */ - if (AH->mode == archModeAppend) - { - if (fn >= 0) - AH->OF = fdopen(dup(fn), PG_BINARY_A); - else - AH->OF = fopen(filename, PG_BINARY_A); - } - else - { - if (fn >= 0) - AH->OF = fdopen(dup(fn), PG_BINARY_W); - else - AH->OF = fopen(filename, PG_BINARY_W); - } - AH->gzOut = 0; - } + AH->OF = cfopen(filename, mode, compression_spec); if (!AH->OF) { @@ -1565,33 +1538,24 @@ SetOutput(ArchiveHandle *AH, const char *filename, } } -static OutputContext +static cfp * SaveOutput(ArchiveHandle *AH) { - OutputContext sav; - - sav.OF = AH->OF; - sav.gzOut = AH->gzOut; - - return sav; + return (cfp *) AH->OF; } static void -RestoreOutput(ArchiveHandle *AH, OutputContext savedContext) +RestoreOutput(ArchiveHandle *AH, cfp *savedOutput) { int res; - errno = 0; /* in case gzclose() doesn't set it */ - if (AH->gzOut) - res = GZCLOSE(AH->OF); - else - res = fclose(AH->OF); + errno = 0; + res = cfclose(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); - AH->gzOut = savedContext.gzOut; - AH->OF = savedContext.OF; + AH->OF = savedOutput; } @@ -1715,22 +1679,17 @@ ahwrite(const void *ptr, size_t size, size_t nmemb, ArchiveHandle *AH) bytes_written = size * nmemb; } - else if (AH->gzOut) - bytes_written = GZWRITE(ptr, size, nmemb, AH->OF); else if (AH->CustomOutPtr) bytes_written = AH->CustomOutPtr(AH, ptr, size * nmemb); + /* + * If we're doing a restore, and it's direct to DB, and we're connected + * then send it to the DB. + */ + else if (RestoringToDB(AH)) + bytes_written = ExecuteSqlCommandBuf(&AH->public, (const char *) ptr, size * nmemb); else - { - /* - * If we're doing a restore, and it's direct to DB, and we're - * connected then send it to the DB. - */ - if (RestoringToDB(AH)) - bytes_written = ExecuteSqlCommandBuf(&AH->public, (const char *) ptr, size * nmemb); - else - bytes_written = fwrite(ptr, size, nmemb, AH->OF) * size; - } + bytes_written = cfwrite(ptr, size * nmemb, AH->OF); if (bytes_written != size * nmemb) WRITE_ERROR_EXIT; @@ -2219,6 +2178,7 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, SetupWorkerPtrType setupWorkerPtr) { ArchiveHandle *AH; + pg_compress_specification out_compress_spec = {0}; pg_log_debug("allocating AH for %s, format %d", FileSpec ? FileSpec : "(stdio)", fmt); @@ -2272,8 +2232,8 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, memset(&(AH->sqlparse), 0, sizeof(AH->sqlparse)); /* Open stdout with no compression for AH output handle */ - AH->gzOut = 0; - AH->OF = stdout; + out_compress_spec.algorithm = PG_COMPRESSION_NONE; + AH->OF = cfdopen(dup(fileno(stdout)), PG_BINARY_A, &out_compress_spec); /* * On Windows, we need to use binary mode to read/write non-text files, diff --git a/src/bin/pg_dump/pg_backup_archiver.h b/src/bin/pg_dump/pg_backup_archiver.h index f72446ed5b4..4725e49747b 100644 --- a/src/bin/pg_dump/pg_backup_archiver.h +++ b/src/bin/pg_dump/pg_backup_archiver.h @@ -32,30 +32,6 @@ #define LOBBUFSIZE 16384 -#ifdef HAVE_LIBZ -#include <zlib.h> -#define GZCLOSE(fh) gzclose(fh) -#define GZWRITE(p, s, n, fh) gzwrite(fh, p, (n) * (s)) -#define GZREAD(p, s, n, fh) gzread(fh, p, (n) * (s)) -#define GZEOF(fh) gzeof(fh) -#else -#define GZCLOSE(fh) fclose(fh) -#define GZWRITE(p, s, n, fh) (fwrite(p, s, n, fh) * (s)) -#define GZREAD(p, s, n, fh) fread(p, s, n, fh) -#define GZEOF(fh) feof(fh) -/* this is just the redefinition of a libz constant */ -#define Z_DEFAULT_COMPRESSION (-1) - -typedef struct _z_stream -{ - void *next_in; - void *next_out; - size_t avail_in; - size_t avail_out; -} z_stream; -typedef z_stream *z_streamp; -#endif - /* Data block types */ #define BLK_DATA 1 #define BLK_BLOBS 3 @@ -319,8 +295,7 @@ struct _archiveHandle char *fSpec; /* Archive File Spec */ FILE *FH; /* General purpose file handle */ - void *OF; - int gzOut; /* Output file */ + void *OF; /* Output file */ struct _tocEntry *toc; /* Header of circular list of TOC entries */ int tocCount; /* Number of TOC entries */ -- 2.25.1 [text/x-diff] 0003-Introduce-Compressor-API-in-pg_dump.patch (65.5K, ../../[email protected]/4-0003-Introduce-Compressor-API-in-pg_dump.patch) download | inline diff: From 6bcd16aa5d94b2fee99ed34fc8a76a757f569cb6 Mon Sep 17 00:00:00 2001 From: Georgios Kokolatos <[email protected]> Date: Wed, 21 Dec 2022 09:49:31 +0000 Subject: [PATCH 3/7] Introduce Compressor API in pg_dump The purpose of this API is to allow for easier addition of new compression methods. CompressFileHandle is substituting the cfp* family of functions under a struct of function pointers for opening, writing, etc. The implementor of a new compression method is now able to "simply" just add those definitions. Custom compressed archives now need to store the compression algorithm in their header. This requires a bump in the version number. The level of compression is no longer stored in the dump as it is irrelevant. --- src/bin/pg_dump/Makefile | 1 + src/bin/pg_dump/compress_gzip.c | 405 ++++++++++++ src/bin/pg_dump/compress_gzip.h | 22 + src/bin/pg_dump/compress_io.c | 916 +++++++------------------- src/bin/pg_dump/compress_io.h | 71 +- src/bin/pg_dump/meson.build | 1 + src/bin/pg_dump/pg_backup_archiver.c | 102 +-- src/bin/pg_dump/pg_backup_archiver.h | 5 +- src/bin/pg_dump/pg_backup_custom.c | 23 +- src/bin/pg_dump/pg_backup_directory.c | 94 +-- src/bin/pg_dump/t/002_pg_dump.pl | 10 +- src/tools/pginclude/cpluspluscheck | 1 + src/tools/pgindent/typedefs.list | 2 + 13 files changed, 852 insertions(+), 801 deletions(-) create mode 100644 src/bin/pg_dump/compress_gzip.c create mode 100644 src/bin/pg_dump/compress_gzip.h diff --git a/src/bin/pg_dump/Makefile b/src/bin/pg_dump/Makefile index ef1ed0f3e51..7a19f5d6172 100644 --- a/src/bin/pg_dump/Makefile +++ b/src/bin/pg_dump/Makefile @@ -24,6 +24,7 @@ LDFLAGS_INTERNAL += -L$(top_builddir)/src/fe_utils -lpgfeutils $(libpq_pgport) OBJS = \ $(WIN32RES) \ + compress_gzip.o \ compress_io.o \ dumputils.o \ parallel.o \ diff --git a/src/bin/pg_dump/compress_gzip.c b/src/bin/pg_dump/compress_gzip.c new file mode 100644 index 00000000000..37c841c5a9b --- /dev/null +++ b/src/bin/pg_dump/compress_gzip.c @@ -0,0 +1,405 @@ +/*------------------------------------------------------------------------- + * + * compress_gzip.c + * Routines for archivers to write an uncompressed or compressed data + * stream. + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_gzip.c + * + *------------------------------------------------------------------------- + */ +#include "postgres_fe.h" +#include <unistd.h> + +#include "compress_gzip.h" +#include "pg_backup_utils.h" + +#ifdef HAVE_LIBZ +#include "zlib.h" + +/*---------------------- + * Compressor API + *---------------------- + */ +typedef struct GzipCompressorState +{ + z_streamp zp; + + void *outbuf; + size_t outsize; +} GzipCompressorState; + +/* Private routines that support gzip compressed data I/O */ +static void +DeflateCompressorGzip(ArchiveHandle *AH, CompressorState *cs, bool flush) +{ + GzipCompressorState *gzipcs = (GzipCompressorState *) cs->private_data; + z_streamp zp = gzipcs->zp; + void *out = gzipcs->outbuf; + int res = Z_OK; + + while (gzipcs->zp->avail_in != 0 || flush) + { + res = deflate(zp, flush ? Z_FINISH : Z_NO_FLUSH); + if (res == Z_STREAM_ERROR) + pg_fatal("could not compress data: %s", zp->msg); + if ((flush && (zp->avail_out < gzipcs->outsize)) + || (zp->avail_out == 0) + || (zp->avail_in != 0) + ) + { + /* + * Extra paranoia: avoid zero-length chunks, since a zero length + * chunk is the EOF marker in the custom format. This should never + * happen but... + */ + if (zp->avail_out < gzipcs->outsize) + { + /* + * Any write function should do its own error checking but to + * make sure we do a check here as well... + */ + size_t len = gzipcs->outsize - zp->avail_out; + + cs->writeF(AH, (char *) out, len); + } + zp->next_out = out; + zp->avail_out = gzipcs->outsize; + } + + if (res == Z_STREAM_END) + break; + } +} + +static void +EndCompressorGzip(ArchiveHandle *AH, CompressorState *cs) +{ + GzipCompressorState *gzipcs = (GzipCompressorState *) cs->private_data; + z_streamp zp; + + if (gzipcs->zp) + { + zp = gzipcs->zp; + zp->next_in = NULL; + zp->avail_in = 0; + + /* Flush any remaining data from zlib buffer */ + DeflateCompressorGzip(AH, cs, true); + + if (deflateEnd(zp) != Z_OK) + pg_fatal("could not close compression stream: %s", zp->msg); + + pg_free(gzipcs->outbuf); + pg_free(gzipcs->zp); + } + + pg_free(gzipcs); + cs->private_data = NULL; +} + +static void +WriteDataToArchiveGzip(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen) +{ + GzipCompressorState *gzipcs = (GzipCompressorState *) cs->private_data; + z_streamp zp; + + if (!gzipcs->zp) + { + zp = gzipcs->zp = (z_streamp) pg_malloc(sizeof(z_stream)); + zp->zalloc = Z_NULL; + zp->zfree = Z_NULL; + zp->opaque = Z_NULL; + + /* + * outsize is the buffer size we tell zlib it can output to. We + * actually allocate one extra byte because some routines want to + * append a trailing zero byte to the zlib output. + */ + gzipcs->outbuf = pg_malloc(ZLIB_OUT_SIZE + 1); + gzipcs->outsize = ZLIB_OUT_SIZE; + + /* + * A level of zero simply copies the input one block at the time. This + * is probably not what the user wanted when calling this interface. + */ + if (cs->compression_spec.level == 0) + pg_fatal("requested to compress the archive yet no level was specified"); + + if (deflateInit(zp, cs->compression_spec.level) != Z_OK) + pg_fatal("could not initialize compression library: %s", zp->msg); + + /* Just be paranoid - maybe End is called after Start, with no Write */ + zp->next_out = gzipcs->outbuf; + zp->avail_out = gzipcs->outsize; + } + + gzipcs->zp->next_in = (void *) unconstify(void *, data); + gzipcs->zp->avail_in = dLen; + DeflateCompressorGzip(AH, cs, false); +} + +static void +ReadDataFromArchiveGzip(ArchiveHandle *AH, CompressorState *cs) +{ + z_streamp zp; + char *out; + int res = Z_OK; + size_t cnt; + char *buf; + size_t buflen; + + zp = (z_streamp) pg_malloc(sizeof(z_stream)); + zp->zalloc = Z_NULL; + zp->zfree = Z_NULL; + zp->opaque = Z_NULL; + + buf = pg_malloc(ZLIB_IN_SIZE); + buflen = ZLIB_IN_SIZE; + + out = pg_malloc(ZLIB_OUT_SIZE + 1); + + if (inflateInit(zp) != Z_OK) + pg_fatal("could not initialize compression library: %s", + zp->msg); + + /* no minimal chunk size for zlib */ + while ((cnt = cs->readF(AH, &buf, &buflen))) + { + zp->next_in = (void *) buf; + zp->avail_in = cnt; + + while (zp->avail_in > 0) + { + zp->next_out = (void *) out; + zp->avail_out = ZLIB_OUT_SIZE; + + res = inflate(zp, 0); + if (res != Z_OK && res != Z_STREAM_END) + pg_fatal("could not uncompress data: %s", zp->msg); + + out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; + ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); + } + } + + zp->next_in = NULL; + zp->avail_in = 0; + while (res != Z_STREAM_END) + { + zp->next_out = (void *) out; + zp->avail_out = ZLIB_OUT_SIZE; + res = inflate(zp, 0); + if (res != Z_OK && res != Z_STREAM_END) + pg_fatal("could not uncompress data: %s", zp->msg); + + out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; + ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); + } + + if (inflateEnd(zp) != Z_OK) + pg_fatal("could not close compression library: %s", zp->msg); + + free(buf); + free(out); + free(zp); +} + +/* Public routines that support gzip compressed data I/O */ +void +InitCompressorGzip(CompressorState *cs, const pg_compress_specification *compression_spec) +{ + GzipCompressorState *gzipcs; + + cs->readData = ReadDataFromArchiveGzip; + cs->writeData = WriteDataToArchiveGzip; + cs->end = EndCompressorGzip; + + cs->compression_spec = *compression_spec; + + gzipcs = (GzipCompressorState *) pg_malloc0(sizeof(GzipCompressorState)); + + cs->private_data = gzipcs; +} + + +/*---------------------- + * Compress File API + *---------------------- + */ + +static size_t +Gzip_read(void *ptr, size_t size, CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + size_t ret; + + ret = gzread(gzfp, ptr, size); + if (ret != size && !gzeof(gzfp)) + { + int errnum; + const char *errmsg = gzerror(gzfp, &errnum); + + pg_fatal("could not read from input file: %s", + errnum == Z_ERRNO ? strerror(errno) : errmsg); + } + + return ret; +} + +static size_t +Gzip_write(const void *ptr, size_t size, CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + + return gzwrite(gzfp, ptr, size); +} + +static int +Gzip_getc(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + int ret; + + errno = 0; + ret = gzgetc(gzfp); + if (ret == EOF) + { + if (!gzeof(gzfp)) + pg_fatal("could not read from input file: %s", strerror(errno)); + else + pg_fatal("could not read from input file: end of file"); + } + + return ret; +} + +static char * +Gzip_gets(char *ptr, int size, CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + + return gzgets(gzfp, ptr, size); +} + +static int +Gzip_close(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + int save_errno; + int ret; + + CFH->private_data = NULL; + + ret = gzclose(gzfp); + + save_errno = errno; + errno = save_errno; + + return ret; +} + +static int +Gzip_eof(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + + return gzeof(gzfp); +} + +static const char * +Gzip_get_error(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + const char *errmsg; + int errnum; + + errmsg = gzerror(gzfp, &errnum); + if (errnum == Z_ERRNO) + errmsg = strerror(errno); + + return errmsg; +} + +static int +Gzip_open(const char *path, int fd, const char *mode, CompressFileHandle *CFH) +{ + gzFile gzfp; + char mode_compression[32]; + + if (CFH->compression_spec.level != Z_DEFAULT_COMPRESSION) + { + /* + * user has specified a compression level, so tell zlib to use it + */ + snprintf(mode_compression, sizeof(mode_compression), "%s%d", + mode, CFH->compression_spec.level); + } + else + strcpy(mode_compression, mode); + + if (fd >= 0) + gzfp = gzdopen(dup(fd), mode_compression); + else + gzfp = gzopen(path, mode_compression); + + if (gzfp == NULL) + return 1; + + CFH->private_data = gzfp; + + return 0; +} + +static int +Gzip_open_write(const char *path, const char *mode, CompressFileHandle *CFH) +{ + char *fname; + int ret; + int save_errno; + + fname = psprintf("%s.gz", path); + ret = CFH->open_func(fname, -1, mode, CFH); + + save_errno = errno; + pg_free(fname); + errno = save_errno; + + return ret; +} + +void +InitCompressGzip(CompressFileHandle *CFH, const pg_compress_specification *compression_spec) +{ + CFH->open_func = Gzip_open; + CFH->open_write_func = Gzip_open_write; + CFH->read_func = Gzip_read; + CFH->write_func = Gzip_write; + CFH->gets_func = Gzip_gets; + CFH->getc_func = Gzip_getc; + CFH->close_func = Gzip_close; + CFH->eof_func = Gzip_eof; + CFH->get_error_func = Gzip_get_error; + + CFH->compression_spec = *compression_spec; + + CFH->private_data = NULL; +} +#else /* HAVE_LIBZ */ +void +InitCompressorGzip(CompressorState *cs, const pg_compress_specification *compression_spec) +{ + pg_fatal("this build does not support compression with %s", "gzip"); +} + +void +InitCompressGzip(CompressFileHandle *CFH, const pg_compress_specification *compression_spec) +{ + pg_fatal("this build does not support compression with %s", "gzip"); +} +#endif /* HAVE_LIBZ */ diff --git a/src/bin/pg_dump/compress_gzip.h b/src/bin/pg_dump/compress_gzip.h new file mode 100644 index 00000000000..a1fc3595e51 --- /dev/null +++ b/src/bin/pg_dump/compress_gzip.h @@ -0,0 +1,22 @@ +/*------------------------------------------------------------------------- + * + * compress_gzip.h + * Interface to compress_io.c routines + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_gzip.h + * + *------------------------------------------------------------------------- + */ +#ifndef _COMPRESS_GZIP_H_ +#define _COMPRESS_GZIP_H_ + +#include "compress_io.h" + +extern void InitCompressorGzip(CompressorState *cs, const pg_compress_specification *compression_spec); +extern void InitCompressGzip(CompressFileHandle *CFH, const pg_compress_specification *compression_spec); + +#endif /* _COMPRESS_GZIP_H_ */ diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index 97b18337578..576a8653193 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -9,42 +9,44 @@ * * This file includes two APIs for dealing with compressed data. The first * provides more flexibility, using callbacks to read/write data from the - * underlying stream. The second API is a wrapper around fopen/gzopen and + * underlying stream. The second API is a wrapper around fopen and * friends, providing an interface similar to those, but abstracts away - * the possible compression. Both APIs use libz for the compression, but - * the second API uses gzip headers, so the resulting files can be easily - * manipulated with the gzip utility. + * the possible compression. The second API is aimed for the resulting + * files can be easily manipulated with an external compression utility + * program. * * Compressor API * -------------- * * The interface for writing to an archive consists of three functions: - * AllocateCompressor, WriteDataToArchive and EndCompressor. First you call - * AllocateCompressor, then write all the data by calling WriteDataToArchive - * as many times as needed, and finally EndCompressor. WriteDataToArchive - * and EndCompressor will call the WriteFunc that was provided to - * AllocateCompressor for each chunk of compressed data. + * AllocateCompressor, writeData, and EndCompressor. First you call + * AllocateCompressor, then write all the data by calling writeData as many + * times as needed, and finally EndCompressor. writeData will call the + * WriteFunc that was provided to AllocateCompressor for each chunk of + * compressed data. * - * The interface for reading an archive consists of just one function: - * ReadDataFromArchive. ReadDataFromArchive reads the whole compressed input - * stream, by repeatedly calling the given ReadFunc. ReadFunc returns the - * compressed data chunk at a time, and ReadDataFromArchive decompresses it - * and passes the decompressed data to ahwrite(), until ReadFunc returns 0 - * to signal EOF. - * - * The interface is the same for compressed and uncompressed streams. + * The interface for reading an archive consists of the same three functions: + * AllocateCompressor, readData, and EndCompressor. First you call + * AllocateCompressor, then read all the data by calling readData to read the + * whole compressed stream which repeatedly calls the given ReadFunc. ReadFunc + * returns the compressed data chunk at a time, and readData decompresses it + * and passes the decompressed data to ahwrite(), until ReadFunc returns 0 to + * signal EOF. The interface is the same for compressed and uncompressed + * streams. * * Compressed stream API * ---------------------- * * The compressed stream API is a wrapper around the C standard fopen() and - * libz's gzopen() APIs. It allows you to use the same functions for - * compressed and uncompressed streams. cfopen_read() first tries to open - * the file with given name, and if it fails, it tries to open the same - * file with the .gz suffix. cfopen_write() opens a file for writing, an - * extra argument specifies if the file should be compressed, and adds the - * .gz suffix to the filename if so. This allows you to easily handle both - * compressed and uncompressed files. + * libz's gzopen() APIs and custom LZ4 calls which provide similar + * functionality. It allows you to use the same functions for compressed and + * uncompressed streams. cfopen_read() first tries to open the file with given + * name, and if it fails, it tries to open the same file with the .gz suffix, + * failing that it tries to open the same file with the .lz4 suffix. + * cfopen_write() opens a file for writing, an extra argument specifies the + * method to use should the file be compressed, and adds the appropriate + * suffix, .gz or .lz4, to the filename if so. This allows you to easily handle + * both compressed and uncompressed files. * * IDENTIFICATION * src/bin/pg_dump/compress_io.c @@ -53,7 +55,11 @@ */ #include "postgres_fe.h" +#include <sys/stat.h> +#include <unistd.h> + #include "compress_io.h" +#include "compress_gzip.h" #include "pg_backup_utils.h" #ifdef HAVE_LIBZ @@ -65,84 +71,70 @@ *---------------------- */ -/* typedef appears in compress_io.h */ -struct CompressorState +static void +ReadDataFromArchiveNone(ArchiveHandle *AH, CompressorState *cs) { - pg_compress_specification compression_spec; - WriteFunc writeF; + size_t cnt; + char *buf; + size_t buflen; -#ifdef HAVE_LIBZ - z_streamp zp; - char *zlibOut; - size_t zlibOutSize; -#endif -}; + buf = pg_malloc(ZLIB_OUT_SIZE); + buflen = ZLIB_OUT_SIZE; + + while ((cnt = cs->readF(AH, &buf, &buflen))) + { + ahwrite(buf, 1, cnt, AH); + } + + free(buf); +} -/* Routines that support zlib compressed data I/O */ -#ifdef HAVE_LIBZ -static void InitCompressorZlib(CompressorState *cs, int level); -static void DeflateCompressorZlib(ArchiveHandle *AH, CompressorState *cs, - bool flush); -static void ReadDataFromArchiveZlib(ArchiveHandle *AH, ReadFunc readF); -static void WriteDataToArchiveZlib(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen); -static void EndCompressorZlib(ArchiveHandle *AH, CompressorState *cs); -#endif -/* Routines that support uncompressed data I/O */ -static void ReadDataFromArchiveNone(ArchiveHandle *AH, ReadFunc readF); -static void WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen); +static void +WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen) +{ + cs->writeF(AH, data, dLen); +} + +static void +EndCompressorNone(ArchiveHandle *AH, CompressorState *cs) +{ + /* no op */ +} + +static void +InitCompressorNone(CompressorState *cs, + const pg_compress_specification *compression_spec) +{ + cs->readData = ReadDataFromArchiveNone; + cs->writeData = WriteDataToArchiveNone; + cs->end = EndCompressorNone; + + cs->compression_spec = *compression_spec; +} /* Public interface routines */ /* Allocate a new compressor */ CompressorState * AllocateCompressor(const pg_compress_specification *compression_spec, - WriteFunc writeF) + ReadFunc readF, WriteFunc writeF) { CompressorState *cs; -#ifndef HAVE_LIBZ - if (compression_spec->algorithm == PG_COMPRESSION_GZIP) - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - cs = (CompressorState *) pg_malloc0(sizeof(CompressorState)); + cs->readF = readF; cs->writeF = writeF; cs->compression_spec = *compression_spec; // XXX - /* - * Perform compression algorithm specific initialization. - */ -#ifdef HAVE_LIBZ - if (cs->compression_spec.algorithm == PG_COMPRESSION_GZIP) - InitCompressorZlib(cs, cs->compression_spec.level); -#endif - - return cs; -} - -/* - * Read all compressed data from the input stream (via readF) and print it - * out with ahwrite(). - */ -void -ReadDataFromArchive(ArchiveHandle *AH, - const pg_compress_specification *compression_spec, - ReadFunc readF) -{ switch (compression_spec->algorithm) { case PG_COMPRESSION_NONE: - ReadDataFromArchiveNone(AH, readF); + InitCompressorNone(cs, compression_spec); break; case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ReadDataFromArchiveZlib(AH, readF); -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif + InitCompressorGzip(cs, compression_spec); break; case PG_COMPRESSION_LZ4: pg_fatal("compression with %s is not yet supported", "LZ4"); @@ -151,34 +143,8 @@ ReadDataFromArchive(ArchiveHandle *AH, pg_fatal("compression with %s is not yet supported", "ZSTD"); break; } -} -/* - * Compress and write data to the output stream (via writeF). - */ -void -WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, - const void *data, size_t dLen) -{ - switch (cs->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - WriteDataToArchiveNone(AH, cs, data, dLen); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - WriteDataToArchiveZlib(AH, cs, data, dLen); -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } + return cs; } /* @@ -187,403 +153,177 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, void EndCompressor(ArchiveHandle *AH, CompressorState *cs) { - switch (cs->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - EndCompressorZlib(AH, cs); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - free(cs); + cs->end(AH, cs); + pg_free(cs); } -/* Private routines, specific to each compression method. */ - -#ifdef HAVE_LIBZ -/* - * Functions for zlib compressed output. +/*---------------------- + * Compressed stream API + *---------------------- */ -static void -InitCompressorZlib(CompressorState *cs, int level) +static int +hasSuffix(const char *filename, const char *suffix) { - z_streamp zp; - - zp = cs->zp = (z_streamp) pg_malloc(sizeof(z_stream)); - zp->zalloc = Z_NULL; - zp->zfree = Z_NULL; - zp->opaque = Z_NULL; - - /* - * zlibOutSize is the buffer size we tell zlib it can output to. We - * actually allocate one extra byte because some routines want to append a - * trailing zero byte to the zlib output. - */ - cs->zlibOut = (char *) pg_malloc(ZLIB_OUT_SIZE + 1); - cs->zlibOutSize = ZLIB_OUT_SIZE; - - if (deflateInit(zp, level) != Z_OK) - pg_fatal("could not initialize compression library: %s", - zp->msg); - - /* Just be paranoid - maybe End is called after Start, with no Write */ - zp->next_out = (void *) cs->zlibOut; - zp->avail_out = cs->zlibOutSize; + int filenamelen = strlen(filename); + int suffixlen = strlen(suffix); + + if (filenamelen < suffixlen) + return 0; + + return memcmp(&filename[filenamelen - suffixlen], + suffix, + suffixlen) == 0; } +/* free() without changing errno; useful in several places below */ static void -EndCompressorZlib(ArchiveHandle *AH, CompressorState *cs) +free_keep_errno(void *p) { - z_streamp zp = cs->zp; - - zp->next_in = NULL; - zp->avail_in = 0; - - /* Flush any remaining data from zlib buffer */ - DeflateCompressorZlib(AH, cs, true); - - if (deflateEnd(zp) != Z_OK) - pg_fatal("could not close compression stream: %s", zp->msg); + int save_errno = errno; - free(cs->zlibOut); - free(cs->zp); + free(p); + errno = save_errno; } -static void -DeflateCompressorZlib(ArchiveHandle *AH, CompressorState *cs, bool flush) +/* + * Compression None implementation + */ +static size_t +read_none(void *ptr, size_t size, CompressFileHandle *CFH) { - z_streamp zp = cs->zp; - char *out = cs->zlibOut; - int res = Z_OK; + FILE *fp = (FILE *) CFH->private_data; + size_t ret; - while (cs->zp->avail_in != 0 || flush) - { - res = deflate(zp, flush ? Z_FINISH : Z_NO_FLUSH); - if (res == Z_STREAM_ERROR) - pg_fatal("could not compress data: %s", zp->msg); - if ((flush && (zp->avail_out < cs->zlibOutSize)) - || (zp->avail_out == 0) - || (zp->avail_in != 0) - ) - { - /* - * Extra paranoia: avoid zero-length chunks, since a zero length - * chunk is the EOF marker in the custom format. This should never - * happen but... - */ - if (zp->avail_out < cs->zlibOutSize) - { - /* - * Any write function should do its own error checking but to - * make sure we do a check here as well... - */ - size_t len = cs->zlibOutSize - zp->avail_out; - - cs->writeF(AH, out, len); - } - zp->next_out = (void *) out; - zp->avail_out = cs->zlibOutSize; - } + if (size == 0) + return 0; - if (res == Z_STREAM_END) - break; - } + ret = fread(ptr, 1, size, fp); + if (ret != size && !feof(fp)) + pg_fatal("could not read from input file: %s", + strerror(errno)); + + return ret; } -static void -WriteDataToArchiveZlib(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen) +static size_t +write_none(const void *ptr, size_t size, CompressFileHandle *CFH) { - cs->zp->next_in = (void *) unconstify(char *, data); - cs->zp->avail_in = dLen; - DeflateCompressorZlib(AH, cs, false); + return fwrite(ptr, 1, size, (FILE *) CFH->private_data); } -static void -ReadDataFromArchiveZlib(ArchiveHandle *AH, ReadFunc readF) +static const char * +get_error_none(CompressFileHandle *CFH) { - z_streamp zp; - char *out; - int res = Z_OK; - size_t cnt; - char *buf; - size_t buflen; - - zp = (z_streamp) pg_malloc(sizeof(z_stream)); - zp->zalloc = Z_NULL; - zp->zfree = Z_NULL; - zp->opaque = Z_NULL; - - buf = pg_malloc(ZLIB_IN_SIZE); - buflen = ZLIB_IN_SIZE; - - out = pg_malloc(ZLIB_OUT_SIZE + 1); - - if (inflateInit(zp) != Z_OK) - pg_fatal("could not initialize compression library: %s", - zp->msg); - - /* no minimal chunk size for zlib */ - while ((cnt = readF(AH, &buf, &buflen))) - { - zp->next_in = (void *) buf; - zp->avail_in = cnt; - - while (zp->avail_in > 0) - { - zp->next_out = (void *) out; - zp->avail_out = ZLIB_OUT_SIZE; + return strerror(errno); +} - res = inflate(zp, 0); - if (res != Z_OK && res != Z_STREAM_END) - pg_fatal("could not uncompress data: %s", zp->msg); +static char * +gets_none(char *ptr, int size, CompressFileHandle *CFH) +{ + return fgets(ptr, size, (FILE *) CFH->private_data); +} - out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; - ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); - } - } +static int +getc_none(CompressFileHandle *CFH) +{ + FILE *fp = (FILE *) CFH->private_data; + int ret; - zp->next_in = NULL; - zp->avail_in = 0; - while (res != Z_STREAM_END) + ret = fgetc(fp); + if (ret == EOF) { - zp->next_out = (void *) out; - zp->avail_out = ZLIB_OUT_SIZE; - res = inflate(zp, 0); - if (res != Z_OK && res != Z_STREAM_END) - pg_fatal("could not uncompress data: %s", zp->msg); - - out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; - ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); + if (!feof(fp)) + pg_fatal("could not read from input file: %s", strerror(errno)); + else + pg_fatal("could not read from input file: end of file"); } - if (inflateEnd(zp) != Z_OK) - pg_fatal("could not close compression library: %s", zp->msg); - - free(buf); - free(out); - free(zp); + return ret; } -#endif /* HAVE_LIBZ */ - - -/* - * Functions for uncompressed output. - */ -static void -ReadDataFromArchiveNone(ArchiveHandle *AH, ReadFunc readF) +static int +close_none(CompressFileHandle *CFH) { - size_t cnt; - char *buf; - size_t buflen; + FILE *fp = (FILE *) CFH->private_data; + int ret = 0; - buf = pg_malloc(ZLIB_OUT_SIZE); - buflen = ZLIB_OUT_SIZE; + CFH->private_data = NULL; - while ((cnt = readF(AH, &buf, &buflen))) - { - ahwrite(buf, 1, cnt, AH); - } + if (fp) + ret = fclose(fp); - free(buf); + return ret; } -static void -WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen) + +static int +eof_none(CompressFileHandle *CFH) { - cs->writeF(AH, data, dLen); + return feof((FILE *) CFH->private_data); } - -/*---------------------- - * Compressed stream API - *---------------------- - */ - -/* - * cfp represents an open stream, wrapping the underlying FILE or gzFile - * pointer. This is opaque to the callers. - */ -struct cfp +static int +open_none(const char *path, int fd, const char *mode, CompressFileHandle *CFH) { - pg_compress_specification compression_spec; - void *fp; -}; + Assert(CFH->private_data == NULL); -#ifdef HAVE_LIBZ -static int hasSuffix(const char *filename, const char *suffix); -#endif + if (fd >= 0) + CFH->private_data = fdopen(dup(fd), mode); + else + CFH->private_data = fopen(path, mode); -/* free() without changing errno; useful in several places below */ -static void -free_keep_errno(void *p) -{ - int save_errno = errno; + if (CFH->private_data == NULL) + return 1; - free(p); - errno = save_errno; + return 0; } -/* - * Open a file for reading. 'path' is the file to open, and 'mode' should - * be either "r" or "rb". - * - * If the file at 'path' does not exist, we append the ".gz" suffix (if 'path' - * doesn't already have it) and try again. So if you pass "foo" as 'path', - * this will open either "foo" or "foo.gz". - * - * On failure, return NULL with an error code in errno. - */ -cfp * -cfopen_read(const char *path, const char *mode) +static int +open_write_none(const char *path, const char *mode, CompressFileHandle *CFH) { - cfp *fp; + Assert(CFH->private_data == NULL); + CFH->private_data = fopen(path, mode); + if (CFH->private_data == NULL) + return 1; - pg_compress_specification compression_spec = {0}; - -#ifdef HAVE_LIBZ - if (hasSuffix(path, ".gz")) - { - compression_spec.algorithm = PG_COMPRESSION_GZIP; - fp = cfopen(path, mode, &compression_spec); - } - else -#endif - { - compression_spec.algorithm = PG_COMPRESSION_NONE; - fp = cfopen(path, mode, &compression_spec); -#ifdef HAVE_LIBZ - if (fp == NULL) - { - char *fname; - - fname = psprintf("%s.gz", path); - compression_spec.algorithm = PG_COMPRESSION_GZIP; - fp = cfopen(fname, mode, &compression_spec); - free_keep_errno(fname); - } -#endif - } - return fp; + return 0; } -/* - * Open a file for writing. 'path' indicates the path name, and 'mode' must - * be a filemode as accepted by fopen() and gzopen() that indicates writing - * ("w", "wb", "a", or "ab"). - * - * If 'compression_spec.algorithm' is GZIP, a gzip compressed stream is opened, - * and 'compression_spec.level' used. The ".gz" suffix is automatically added to - * 'path' in that case. - * - * On failure, return NULL with an error code in errno. - */ -cfp * -cfopen_write(const char *path, const char *mode, - const pg_compress_specification *compression_spec) +static void +InitCompressNone(CompressFileHandle *CFH, + const pg_compress_specification *compression_spec) { - cfp *fp; - - if (compression_spec->algorithm == PG_COMPRESSION_NONE) - fp = cfopen(path, mode, compression_spec); - else - { -#ifdef HAVE_LIBZ - char *fname; - - fname = psprintf("%s.gz", path); - fp = cfopen(fname, mode, compression_spec); - free_keep_errno(fname); -#else - pg_fatal("this build does not support compression with %s", "gzip"); - fp = NULL; /* keep compiler quiet */ -#endif - } - return fp; + CFH->open_func = open_none; + CFH->open_write_func = open_write_none; + CFH->read_func = read_none; + CFH->write_func = write_none; + CFH->gets_func = gets_none; + CFH->getc_func = getc_none; + CFH->close_func = close_none; + CFH->eof_func = eof_none; + CFH->get_error_func = get_error_none; + + CFH->private_data = NULL; } /* - * This is the workhorse for cfopen() or cfdopen(). It opens file 'path' or - * associates a stream 'fd', if 'fd' is a valid descriptor, in 'mode'. The - * descriptor is not dup'ed and it is the caller's responsibility to do so. - * The caller must verify that the 'compress_algorithm' is supported by the - * current build. - * - * On failure, return NULL with an error code in errno. + * Public interface */ -static cfp * -cfopen_internal(const char *path, int fd, const char *mode, - const pg_compress_specification *compression_spec) +CompressFileHandle * +InitCompressFileHandle(const pg_compress_specification *compression_spec) { - cfp *fp = pg_malloc(sizeof(cfp)); + CompressFileHandle *CFH; - fp->compression_spec = *compression_spec; + CFH = pg_malloc0(sizeof(CompressFileHandle)); switch (compression_spec->algorithm) { case PG_COMPRESSION_NONE: - if (fd >= 0) - fp->fp = fdopen(fd, mode); - else - fp->fp = fopen(path, mode); - if (fp->fp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } - + InitCompressNone(CFH, compression_spec); break; case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - if (compression_spec->level != Z_DEFAULT_COMPRESSION) - { - /* - * user has specified a compression level, so tell zlib to use - * it - */ - char mode_compression[32]; - - snprintf(mode_compression, sizeof(mode_compression), "%s%d", - mode, compression_spec->level); - if (fd >= 0) - fp->fp = gzdopen(fd, mode_compression); - // fp->compressedfp = gzopen(path, mode_compression); - else - fp->fp = gzopen(path, mode_compression); - } - else - { - /* don't specify a level, just use the zlib default */ - if (fd >= 0) - fp->fp = gzdopen(fd, mode); - else - fp->fp = gzopen(path, mode); - // fp->compressedfp = gzopen(path, mode); - } - - if (fp->fp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif + InitCompressGzip(CFH, compression_spec); break; case PG_COMPRESSION_LZ4: pg_fatal("compression with %s is not yet supported", "LZ4"); @@ -593,266 +333,88 @@ cfopen_internal(const char *path, int fd, const char *mode, break; } - return fp; + return CFH; } -cfp * -cfopen(const char *path, const char *mode, - const pg_compress_specification *compression_spec) +/* + * Open a file for reading. 'path' is the file to open, and 'mode' should + * be either "r" or "rb". + * + * If the file at 'path' does not exist, we append the "{.gz,.lz4}" suffix (i + * 'path' doesn't already have it) and try again. So if you pass "foo" as + * 'path', this will open either "foo" or "foo.gz" or "foo.lz4", trying in that + * order. + * + * On failure, return NULL with an error code in errno. + */ +CompressFileHandle * +InitDiscoverCompressFileHandle(const char *path, const char *mode) { - return cfopen_internal(path, -1, mode, compression_spec); -} + CompressFileHandle *CFH = NULL; + struct stat st; + char *fname; + pg_compress_specification compression_spec = {0}; -cfp * -cfdopen(int fd, const char *mode, - const pg_compress_specification *compression_spec) -{ - return cfopen_internal(NULL, fd, mode, compression_spec); -} + // compression_spec.algorithm = PG_COMPRESSION_NONE; -int -cfread(void *ptr, int size, cfp *fp) -{ - int ret = 0; + Assert(strcmp(mode, "r") == 0 || strcmp(mode, "rb") == 0); - if (size == 0) - return 0; + fname = strdup(path); - switch (fp->compression_spec.algorithm) + if (hasSuffix(fname, ".gz")) + compression_spec.algorithm = PG_COMPRESSION_GZIP; + else { - case PG_COMPRESSION_NONE: - ret = fread(ptr, 1, size, (FILE *) fp->fp); - if (ret != size && !feof((FILE *) fp->fp)) - READ_ERROR_EXIT((FILE *) fp->fp); + bool exists; - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzread((gzFile) fp->fp, ptr, size); - if (ret != size && !gzeof((gzFile) fp->fp)) - { - int errnum; - const char *errmsg = gzerror((gzFile) fp->fp, &errnum); - - pg_fatal("could not read from input file: %s", - errnum == Z_ERRNO ? strerror(errno) : errmsg); - } -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - return ret; -} - -int -cfwrite(const void *ptr, int size, cfp *fp) -{ - int ret = 0; - - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = fwrite(ptr, 1, size, (FILE *) fp->fp); - break; - case PG_COMPRESSION_GZIP: + exists = (stat(path, &st) == 0); + /* avoid unused warning if it is not build with compression */ + if (exists) + compression_spec.algorithm = PG_COMPRESSION_NONE; #ifdef HAVE_LIBZ - ret = gzwrite((gzFile) fp->fp, ptr, size); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - return ret; -} - -int -cfgetc(cfp *fp) -{ - int ret = 0; - - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = fgetc((FILE *) fp->fp); - if (ret == EOF) - READ_ERROR_EXIT((FILE *) fp->fp); + if (!exists) + { + free_keep_errno(fname); + fname = psprintf("%s.gz", path); + exists = (stat(fname, &st) == 0); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzgetc((gzFile) fp->fp); - if (ret == EOF) - { - if (!gzeof((gzFile) fp->fp)) - pg_fatal("could not read from input file: %s", strerror(errno)); - else - pg_fatal("could not read from input file: end of file"); - } -#else - pg_fatal("this build does not support compression with %s", - "gzip"); + if (exists) + compression_spec.algorithm = PG_COMPRESSION_GZIP; + } #endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - return ret; -} - -char * -cfgets(cfp *fp, char *buf, int len) -{ - char *ret = NULL; - - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = fgets(buf, len, (FILE *) fp->fp); +#ifdef USE_LZ4 + if (!exists) + { + free_keep_errno(fname); + fname = psprintf("%s.lz4", path); + exists = (stat(fname, &st) == 0); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzgets((gzFile) fp->fp, buf, len); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); + if (exists) + compression_spec.algorithm = PG_COMPRESSION_LZ4; + } #endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; } - return ret; -} - -int -cfclose(cfp *fp) -{ - int ret = 0; - - if (fp == NULL) + CFH = InitCompressFileHandle(&compression_spec); + if (CFH->open_func(fname, -1, mode, CFH)) { - errno = EBADF; - return EOF; + free_keep_errno(CFH); + CFH = NULL; } + free_keep_errno(fname); - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = fclose((FILE *) fp->fp); - fp->fp = NULL; - - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzclose((gzFile) fp->fp); - fp->fp = NULL; -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - free_keep_errno(fp); - - return ret; + return CFH; } int -cfeof(cfp *fp) +DestroyCompressFileHandle(CompressFileHandle *CFH) { int ret = 0; - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = feof((FILE *) fp->fp); + if (CFH->private_data) + ret = CFH->close_func(CFH); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzeof((gzFile) fp->fp); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } + free_keep_errno(CFH); return ret; } - -const char * -get_cfp_error(cfp *fp) -{ - if (fp->compression_spec.algorithm == PG_COMPRESSION_GZIP) - { -#ifdef HAVE_LIBZ - int errnum; - const char *errmsg = gzerror((gzFile) fp->fp, &errnum); - - if (errnum != Z_ERRNO) - return errmsg; -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - } - - return strerror(errno); -} - -#ifdef HAVE_LIBZ -static int -hasSuffix(const char *filename, const char *suffix) -{ - int filenamelen = strlen(filename); - int suffixlen = strlen(suffix); - - if (filenamelen < suffixlen) - return 0; - - return memcmp(&filename[filenamelen - suffixlen], - suffix, - suffixlen) == 0; -} - -#endif diff --git a/src/bin/pg_dump/compress_io.h b/src/bin/pg_dump/compress_io.h index 768096c820d..afe6b22efaf 100644 --- a/src/bin/pg_dump/compress_io.h +++ b/src/bin/pg_dump/compress_io.h @@ -37,34 +37,63 @@ typedef void (*WriteFunc) (ArchiveHandle *AH, const char *buf, size_t len); */ typedef size_t (*ReadFunc) (ArchiveHandle *AH, char **buf, size_t *buflen); -/* struct definition appears in compress_io.c */ typedef struct CompressorState CompressorState; +struct CompressorState +{ + /* + * Read all compressed data from the input stream (via readF) and print it + * out with ahwrite(). + */ + void (*readData) (ArchiveHandle *AH, CompressorState *cs); + + /* + * Compress and write data to the output stream (via writeF). + */ + void (*writeData) (ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen); + void (*end) (ArchiveHandle *AH, CompressorState *cs); + + ReadFunc readF; + WriteFunc writeF; + + pg_compress_specification compression_spec; + void *private_data; +}; extern CompressorState *AllocateCompressor(const pg_compress_specification *compression_spec, + ReadFunc readF, WriteFunc writeF); -extern void ReadDataFromArchive(ArchiveHandle *AH, - const pg_compress_specification *compression_spec, - ReadFunc readF); -extern void WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, - const void *data, size_t dLen); extern void EndCompressor(ArchiveHandle *AH, CompressorState *cs); +/* + * Compress File Handle + */ +typedef struct CompressFileHandle CompressFileHandle; + +struct CompressFileHandle +{ + int (*open_func) (const char *path, int fd, const char *mode, + CompressFileHandle *CFH); + int (*open_write_func) (const char *path, const char *mode, + CompressFileHandle *cxt); + size_t (*read_func) (void *ptr, size_t size, CompressFileHandle *CFH); + size_t (*write_func) (const void *ptr, size_t size, + struct CompressFileHandle *CFH); + char *(*gets_func) (char *s, int size, CompressFileHandle *CFH); + int (*getc_func) (CompressFileHandle *CFH); + int (*eof_func) (CompressFileHandle *CFH); + int (*close_func) (CompressFileHandle *CFH); + const char *(*get_error_func) (CompressFileHandle *CFH); + + pg_compress_specification compression_spec; + void *private_data; +}; -typedef struct cfp cfp; +extern CompressFileHandle *InitCompressFileHandle( + const pg_compress_specification *compression_spec); -extern cfp *cfopen(const char *path, const char *mode, - const pg_compress_specification *compression_spec); -extern cfp *cfdopen(int fd, const char *mode, - const pg_compress_specification *compression_spec); -extern cfp *cfopen_read(const char *path, const char *mode); -extern cfp *cfopen_write(const char *path, const char *mode, - const pg_compress_specification *compression_spec); -extern int cfread(void *ptr, int size, cfp *fp); -extern int cfwrite(const void *ptr, int size, cfp *fp); -extern int cfgetc(cfp *fp); -extern char *cfgets(cfp *fp, char *buf, int len); -extern int cfclose(cfp *fp); -extern int cfeof(cfp *fp); -extern const char *get_cfp_error(cfp *fp); +extern CompressFileHandle *InitDiscoverCompressFileHandle(const char *path, + const char *mode); +extern int DestroyCompressFileHandle(CompressFileHandle *CFH); #endif diff --git a/src/bin/pg_dump/meson.build b/src/bin/pg_dump/meson.build index ca62f9a3740..aa2c91829c0 100644 --- a/src/bin/pg_dump/meson.build +++ b/src/bin/pg_dump/meson.build @@ -2,6 +2,7 @@ pg_dump_common_sources = files( 'compress_io.c', + 'compress_gzip.c', 'dumputils.c', 'parallel.c', 'pg_backup_archiver.c', diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index cbe110c917a..06f0b46cbfc 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -95,8 +95,8 @@ static void dump_lo_buf(ArchiveHandle *AH); static void dumpTimestamp(ArchiveHandle *AH, const char *msg, time_t tim); static void SetOutput(ArchiveHandle *AH, const char *filename, const pg_compress_specification *compression_spec); -static cfp *SaveOutput(ArchiveHandle *AH); -static void RestoreOutput(ArchiveHandle *AH, cfp *savedOutput); +static CompressFileHandle *SaveOutput(ArchiveHandle *AH); +static void RestoreOutput(ArchiveHandle *AH, CompressFileHandle *savedOutput); static int restore_toc_entry(ArchiveHandle *AH, TocEntry *te, bool is_parallel); static void restore_toc_entries_prefork(ArchiveHandle *AH, @@ -272,7 +272,7 @@ CloseArchive(Archive *AHX) /* Close the output */ errno = 0; - res = cfclose(AH->OF); + res = DestroyCompressFileHandle(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); @@ -355,7 +355,7 @@ RestoreArchive(Archive *AHX) bool parallel_mode; bool supports_compression; TocEntry *te; - cfp *sav; + CompressFileHandle *sav; AH->stage = STAGE_INITIALIZING; @@ -1127,7 +1127,7 @@ PrintTOCSummary(Archive *AHX) TocEntry *te; pg_compress_specification out_compression_spec = {0}; teSection curSection; - cfp *sav; + CompressFileHandle *sav; const char *fmtName; char stamp_str[64]; @@ -1143,9 +1143,10 @@ PrintTOCSummary(Archive *AHX) strcpy(stamp_str, "[unknown]"); ahprintf(AH, ";\n; Archive created at %s\n", stamp_str); - ahprintf(AH, "; dbname: %s\n; TOC Entries: %d\n; Compression: %d\n", + ahprintf(AH, "; dbname: %s\n; TOC Entries: %d\n; Compression: %s\n", sanitize_line(AH->archdbname, false), - AH->tocCount, AH->compression_spec.level); + AH->tocCount, + get_compress_algorithm_name(AH->compression_spec.algorithm)); switch (AH->format) { @@ -1502,6 +1503,7 @@ static void SetOutput(ArchiveHandle *AH, const char *filename, const pg_compress_specification *compression_spec) { + CompressFileHandle *CFH; const char *mode; int fn = -1; @@ -1524,33 +1526,32 @@ SetOutput(ArchiveHandle *AH, const char *filename, else mode = PG_BINARY_W; - if (fn >= 0) - AH->OF = cfdopen(dup(fn), mode, compression_spec); - else - AH->OF = cfopen(filename, mode, compression_spec); + CFH = InitCompressFileHandle(compression_spec); - if (!AH->OF) + if (CFH->open_func(filename, fn, mode, CFH)) { if (filename) pg_fatal("could not open output file \"%s\": %m", filename); else pg_fatal("could not open output file: %m"); } + + AH->OF = CFH; } -static cfp * +static CompressFileHandle * SaveOutput(ArchiveHandle *AH) { - return (cfp *) AH->OF; + return (CompressFileHandle *) AH->OF; } static void -RestoreOutput(ArchiveHandle *AH, cfp *savedOutput) +RestoreOutput(ArchiveHandle *AH, CompressFileHandle *savedOutput) { int res; errno = 0; - res = cfclose(AH->OF); + res = DestroyCompressFileHandle(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); @@ -1689,7 +1690,11 @@ ahwrite(const void *ptr, size_t size, size_t nmemb, ArchiveHandle *AH) else if (RestoringToDB(AH)) bytes_written = ExecuteSqlCommandBuf(&AH->public, (const char *) ptr, size * nmemb); else - bytes_written = cfwrite(ptr, size * nmemb, AH->OF); + { + CompressFileHandle *CFH = (CompressFileHandle *) AH->OF; + + bytes_written = CFH->write_func(ptr, size * nmemb, CFH); + } if (bytes_written != size * nmemb) WRITE_ERROR_EXIT; @@ -2031,6 +2036,18 @@ ReadStr(ArchiveHandle *AH) return buf; } +static bool +_fileExistsInDirectory(const char *dir, const char *filename) +{ + struct stat st; + char buf[MAXPGPATH]; + + if (snprintf(buf, MAXPGPATH, "%s/%s", dir, filename) >= MAXPGPATH) + pg_fatal("directory name too long: \"%s\"", dir); + + return (stat(buf, &st) == 0 && S_ISREG(st.st_mode)); +} + static int _discoverArchiveFormat(ArchiveHandle *AH) { @@ -2061,26 +2078,12 @@ _discoverArchiveFormat(ArchiveHandle *AH) */ if (stat(AH->fSpec, &st) == 0 && S_ISDIR(st.st_mode)) { - char buf[MAXPGPATH]; - - if (snprintf(buf, MAXPGPATH, "%s/toc.dat", AH->fSpec) >= MAXPGPATH) - pg_fatal("directory name too long: \"%s\"", - AH->fSpec); - if (stat(buf, &st) == 0 && S_ISREG(st.st_mode)) - { - AH->format = archDirectory; + AH->format = archDirectory; + if (_fileExistsInDirectory(AH->fSpec, "toc.dat")) return AH->format; - } - #ifdef HAVE_LIBZ - if (snprintf(buf, MAXPGPATH, "%s/toc.dat.gz", AH->fSpec) >= MAXPGPATH) - pg_fatal("directory name too long: \"%s\"", - AH->fSpec); - if (stat(buf, &st) == 0 && S_ISREG(st.st_mode)) - { - AH->format = archDirectory; + if (_fileExistsInDirectory(AH->fSpec, "toc.dat.gz")) return AH->format; - } #endif pg_fatal("directory \"%s\" does not appear to be a valid archive (\"toc.dat\" does not exist)", AH->fSpec); @@ -2178,6 +2181,7 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, SetupWorkerPtrType setupWorkerPtr) { ArchiveHandle *AH; + CompressFileHandle *CFH; pg_compress_specification out_compress_spec = {0}; pg_log_debug("allocating AH for %s, format %d", @@ -2233,7 +2237,10 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, /* Open stdout with no compression for AH output handle */ out_compress_spec.algorithm = PG_COMPRESSION_NONE; - AH->OF = cfdopen(dup(fileno(stdout)), PG_BINARY_A, &out_compress_spec); + CFH = InitCompressFileHandle(&out_compress_spec); + if (CFH->open_func(NULL, fileno(stdout), PG_BINARY_A, CFH)) + pg_fatal("could not open stdout for appending: %m"); + AH->OF = CFH; /* * On Windows, we need to use binary mode to read/write non-text files, @@ -3646,12 +3653,7 @@ WriteHead(ArchiveHandle *AH) AH->WriteBytePtr(AH, AH->intSize); AH->WriteBytePtr(AH, AH->offSize); AH->WriteBytePtr(AH, AH->format); - /* - * For now the compression type is implied by the level. This will need - * to change once support for more compression algorithms is added, - * requiring a format bump. - */ - WriteInt(AH, AH->compression_spec.level); + AH->WriteBytePtr(AH, AH->compression_spec.algorithm); crtm = *localtime(&AH->createDate); WriteInt(AH, crtm.tm_sec); WriteInt(AH, crtm.tm_min); @@ -3722,10 +3724,11 @@ ReadHead(ArchiveHandle *AH) pg_fatal("expected format (%d) differs from format found in file (%d)", AH->format, fmt); - /* Guess the compression method based on the level */ - AH->compression_spec.algorithm = PG_COMPRESSION_NONE; - if (AH->version >= K_VERS_1_2) + if (AH->version >= K_VERS_1_15) + AH->compression_spec.algorithm = AH->ReadBytePtr(AH); + else if (AH->version >= K_VERS_1_2) { + /* Guess the compression method based on the level */ if (AH->version < K_VERS_1_4) AH->compression_spec.level = AH->ReadBytePtr(AH); else @@ -3737,10 +3740,17 @@ ReadHead(ArchiveHandle *AH) else AH->compression_spec.algorithm = PG_COMPRESSION_GZIP; + if (AH->compression_spec.algorithm != PG_COMPRESSION_NONE) + { + bool unsupported = false; + #ifndef HAVE_LIBZ - if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) - pg_fatal("archive is compressed, but this installation does not support compression"); + if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) + unsupported = true; #endif + if (unsupported) + pg_fatal("archive is compressed, but this installation does not support compression"); + } if (AH->version >= K_VERS_1_4) { diff --git a/src/bin/pg_dump/pg_backup_archiver.h b/src/bin/pg_dump/pg_backup_archiver.h index 4725e49747b..18b38c17abc 100644 --- a/src/bin/pg_dump/pg_backup_archiver.h +++ b/src/bin/pg_dump/pg_backup_archiver.h @@ -65,10 +65,13 @@ #define K_VERS_1_13 MAKE_ARCHIVE_VERSION(1, 13, 0) /* change search_path * behavior */ #define K_VERS_1_14 MAKE_ARCHIVE_VERSION(1, 14, 0) /* add tableam */ +#define K_VERS_1_15 MAKE_ARCHIVE_VERSION(1, 15, 0) /* add + * compression_algorithm + * in header */ /* Current archive version number (the format we can output) */ #define K_VERS_MAJOR 1 -#define K_VERS_MINOR 14 +#define K_VERS_MINOR 15 #define K_VERS_REV 0 #define K_VERS_SELF MAKE_ARCHIVE_VERSION(K_VERS_MAJOR, K_VERS_MINOR, K_VERS_REV) diff --git a/src/bin/pg_dump/pg_backup_custom.c b/src/bin/pg_dump/pg_backup_custom.c index 0e87444de85..40cd90b7325 100644 --- a/src/bin/pg_dump/pg_backup_custom.c +++ b/src/bin/pg_dump/pg_backup_custom.c @@ -298,7 +298,9 @@ _StartData(ArchiveHandle *AH, TocEntry *te) _WriteByte(AH, BLK_DATA); /* Block type */ WriteInt(AH, te->dumpId); /* For sanity check */ - ctx->cs = AllocateCompressor(&AH->compression_spec, _CustomWriteFunc); + ctx->cs = AllocateCompressor(&AH->compression_spec, + NULL, + _CustomWriteFunc); } /* @@ -317,15 +319,15 @@ _WriteData(ArchiveHandle *AH, const void *data, size_t dLen) CompressorState *cs = ctx->cs; if (dLen > 0) - /* WriteDataToArchive() internally throws write errors */ - WriteDataToArchive(AH, cs, data, dLen); + /* writeData() internally throws write errors */ + cs->writeData(AH, cs, data, dLen); } /* * Called by the archiver when a dumper's 'DataDumper' routine has * finished. * - * Optional. + * Mandatory. */ static void _EndData(ArchiveHandle *AH, TocEntry *te) @@ -333,6 +335,8 @@ _EndData(ArchiveHandle *AH, TocEntry *te) lclContext *ctx = (lclContext *) AH->formatData; EndCompressor(AH, ctx->cs); + ctx->cs = NULL; + /* Send the end marker */ WriteInt(AH, 0); } @@ -377,7 +381,9 @@ _StartLO(ArchiveHandle *AH, TocEntry *te, Oid oid) WriteInt(AH, oid); - ctx->cs = AllocateCompressor(&AH->compression_spec, _CustomWriteFunc); + ctx->cs = AllocateCompressor(&AH->compression_spec, + NULL, + _CustomWriteFunc); } /* @@ -566,7 +572,12 @@ _PrintTocData(ArchiveHandle *AH, TocEntry *te) static void _PrintData(ArchiveHandle *AH) { - ReadDataFromArchive(AH, &AH->compression_spec, _CustomReadFunc); + CompressorState *cs; + + cs = AllocateCompressor(&AH->compression_spec, + _CustomReadFunc, NULL); + cs->readData(AH, cs); + EndCompressor(AH, cs); } static void diff --git a/src/bin/pg_dump/pg_backup_directory.c b/src/bin/pg_dump/pg_backup_directory.c index ffb8a0e4dd7..2d4baf58c22 100644 --- a/src/bin/pg_dump/pg_backup_directory.c +++ b/src/bin/pg_dump/pg_backup_directory.c @@ -50,9 +50,8 @@ typedef struct */ char *directory; - cfp *dataFH; /* currently open data file */ - - cfp *LOsTocFH; /* file handle for blobs.toc */ + CompressFileHandle *dataFH; /* currently open data file */ + CompressFileHandle *LOsTocFH; /* file handle for blobs.toc */ ParallelState *pstate; /* for parallel backup / restore */ } lclContext; @@ -198,11 +197,11 @@ InitArchiveFmt_Directory(ArchiveHandle *AH) else { /* Read Mode */ char fname[MAXPGPATH]; - cfp *tocFH; + CompressFileHandle *tocFH; setFilePath(AH, fname, "toc.dat"); - tocFH = cfopen_read(fname, PG_BINARY_R); + tocFH = InitDiscoverCompressFileHandle(fname, PG_BINARY_R); if (tocFH == NULL) pg_fatal("could not open input file \"%s\": %m", fname); @@ -218,7 +217,7 @@ InitArchiveFmt_Directory(ArchiveHandle *AH) ReadToc(AH); /* Nothing else in the file, so close it again... */ - if (cfclose(tocFH) != 0) + if (DestroyCompressFileHandle(tocFH) != 0) pg_fatal("could not close TOC file: %m"); ctx->dataFH = NULL; } @@ -327,9 +326,9 @@ _StartData(ArchiveHandle *AH, TocEntry *te) setFilePath(AH, fname, tctx->filename); - ctx->dataFH = cfopen_write(fname, PG_BINARY_W, - &AH->compression_spec); - if (ctx->dataFH == NULL) + ctx->dataFH = InitCompressFileHandle(&AH->compression_spec); + + if (ctx->dataFH->open_write_func(fname, PG_BINARY_W, ctx->dataFH)) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -346,15 +345,16 @@ static void _WriteData(ArchiveHandle *AH, const void *data, size_t dLen) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; errno = 0; - if (dLen > 0 && cfwrite(data, dLen, ctx->dataFH) != dLen) + if (dLen > 0 && CFH->write_func(data, dLen, CFH) != dLen) { /* if write didn't set errno, assume problem is no disk space */ if (errno == 0) errno = ENOSPC; pg_fatal("could not write to output file: %s", - get_cfp_error(ctx->dataFH)); + CFH->get_error_func(CFH)); } } @@ -370,7 +370,7 @@ _EndData(ArchiveHandle *AH, TocEntry *te) lclContext *ctx = (lclContext *) AH->formatData; /* Close the file */ - if (cfclose(ctx->dataFH) != 0) + if (DestroyCompressFileHandle(ctx->dataFH) != 0) pg_fatal("could not close data file: %m"); ctx->dataFH = NULL; @@ -385,26 +385,25 @@ _PrintFileData(ArchiveHandle *AH, char *filename) size_t cnt; char *buf; size_t buflen; - cfp *cfp; + CompressFileHandle *CFH; if (!filename) return; - cfp = cfopen_read(filename, PG_BINARY_R); - - if (!cfp) + CFH = InitDiscoverCompressFileHandle(filename, PG_BINARY_R); + if (!CFH) pg_fatal("could not open input file \"%s\": %m", filename); buf = pg_malloc(ZLIB_OUT_SIZE); buflen = ZLIB_OUT_SIZE; - while ((cnt = cfread(buf, buflen, cfp))) + while ((cnt = CFH->read_func(buf, buflen, CFH))) { ahwrite(buf, 1, cnt, AH); } free(buf); - if (cfclose(cfp) != 0) + if (DestroyCompressFileHandle(CFH) != 0) pg_fatal("could not close data file \"%s\": %m", filename); } @@ -435,6 +434,7 @@ _LoadLOs(ArchiveHandle *AH) { Oid oid; lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH; char tocfname[MAXPGPATH]; char line[MAXPGPATH]; @@ -442,14 +442,14 @@ _LoadLOs(ArchiveHandle *AH) setFilePath(AH, tocfname, "blobs.toc"); - ctx->LOsTocFH = cfopen_read(tocfname, PG_BINARY_R); + CFH = ctx->LOsTocFH = InitDiscoverCompressFileHandle(tocfname, PG_BINARY_R); if (ctx->LOsTocFH == NULL) pg_fatal("could not open large object TOC file \"%s\" for input: %m", tocfname); /* Read the LOs TOC file line-by-line, and process each LO */ - while ((cfgets(ctx->LOsTocFH, line, MAXPGPATH)) != NULL) + while ((CFH->gets_func(line, MAXPGPATH, CFH)) != NULL) { char lofname[MAXPGPATH + 1]; char path[MAXPGPATH]; @@ -464,11 +464,11 @@ _LoadLOs(ArchiveHandle *AH) _PrintFileData(AH, path); EndRestoreLO(AH, oid); } - if (!cfeof(ctx->LOsTocFH)) + if (!CFH->eof_func(CFH)) pg_fatal("error reading large object TOC file \"%s\"", tocfname); - if (cfclose(ctx->LOsTocFH) != 0) + if (DestroyCompressFileHandle(ctx->LOsTocFH) != 0) pg_fatal("could not close large object TOC file \"%s\": %m", tocfname); @@ -488,15 +488,16 @@ _WriteByte(ArchiveHandle *AH, const int i) { unsigned char c = (unsigned char) i; lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; errno = 0; - if (cfwrite(&c, 1, ctx->dataFH) != 1) + if (CFH->write_func(&c, 1, CFH) != 1) { /* if write didn't set errno, assume problem is no disk space */ if (errno == 0) errno = ENOSPC; pg_fatal("could not write to output file: %s", - get_cfp_error(ctx->dataFH)); + CFH->get_error_func(CFH)); } return 1; @@ -512,8 +513,9 @@ static int _ReadByte(ArchiveHandle *AH) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; - return cfgetc(ctx->dataFH); + return CFH->getc_func(CFH); } /* @@ -524,15 +526,16 @@ static void _WriteBuf(ArchiveHandle *AH, const void *buf, size_t len) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; errno = 0; - if (cfwrite(buf, len, ctx->dataFH) != len) + if (CFH->write_func(buf, len, CFH) != len) { /* if write didn't set errno, assume problem is no disk space */ if (errno == 0) errno = ENOSPC; pg_fatal("could not write to output file: %s", - get_cfp_error(ctx->dataFH)); + CFH->get_error_func(CFH)); } } @@ -545,12 +548,13 @@ static void _ReadBuf(ArchiveHandle *AH, void *buf, size_t len) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; /* - * If there was an I/O error, we already exited in cfread(), so here we + * If there was an I/O error, we already exited in readF(), so here we * exit on short reads. */ - if (cfread(buf, len, ctx->dataFH) != len) + if (CFH->read_func(buf, len, CFH) != len) pg_fatal("could not read from input file: end of file"); } @@ -573,7 +577,7 @@ _CloseArchive(ArchiveHandle *AH) if (AH->mode == archModeWrite) { - cfp *tocFH; + CompressFileHandle *tocFH; pg_compress_specification compression_spec = {0}; char fname[MAXPGPATH]; @@ -584,8 +588,8 @@ _CloseArchive(ArchiveHandle *AH) /* The TOC is always created uncompressed */ compression_spec.algorithm = PG_COMPRESSION_NONE; - tocFH = cfopen_write(fname, PG_BINARY_W, &compression_spec); - if (tocFH == NULL) + tocFH = InitCompressFileHandle(&compression_spec); + if (tocFH->open_write_func(fname, PG_BINARY_W, tocFH)) pg_fatal("could not open output file \"%s\": %m", fname); ctx->dataFH = tocFH; @@ -598,7 +602,7 @@ _CloseArchive(ArchiveHandle *AH) WriteHead(AH); AH->format = archDirectory; WriteToc(AH); - if (cfclose(tocFH) != 0) + if (DestroyCompressFileHandle(tocFH) != 0) pg_fatal("could not close TOC file: %m"); WriteDataChunks(AH, ctx->pstate); @@ -649,8 +653,8 @@ _StartLOs(ArchiveHandle *AH, TocEntry *te) /* The LO TOC file is never compressed */ compression_spec.algorithm = PG_COMPRESSION_NONE; - ctx->LOsTocFH = cfopen_write(fname, "ab", &compression_spec); - if (ctx->LOsTocFH == NULL) + ctx->LOsTocFH = InitCompressFileHandle(&compression_spec); + if (ctx->LOsTocFH->open_write_func(fname, "ab", ctx->LOsTocFH)) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -667,9 +671,8 @@ _StartLO(ArchiveHandle *AH, TocEntry *te, Oid oid) snprintf(fname, MAXPGPATH, "%s/blob_%u.dat", ctx->directory, oid); - ctx->dataFH = cfopen_write(fname, PG_BINARY_W, &AH->compression_spec); - - if (ctx->dataFH == NULL) + ctx->dataFH = InitCompressFileHandle(&AH->compression_spec); + if (ctx->dataFH->open_write_func(fname, PG_BINARY_W, ctx->dataFH)) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -682,18 +685,19 @@ static void _EndLO(ArchiveHandle *AH, TocEntry *te, Oid oid) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->LOsTocFH; char buf[50]; int len; - /* Close the LO data file itself */ - if (cfclose(ctx->dataFH) != 0) - pg_fatal("could not close LO data file: %m"); + /* Close the BLOB data file itself */ + if (DestroyCompressFileHandle(ctx->dataFH) != 0) + pg_fatal("could not close blob data file: %m"); ctx->dataFH = NULL; /* register the LO in blobs.toc */ len = snprintf(buf, sizeof(buf), "%u blob_%u.dat\n", oid, oid); - if (cfwrite(buf, len, ctx->LOsTocFH) != len) - pg_fatal("could not write to LOs TOC file"); + if (CFH->write_func(buf, len, CFH) != len) + pg_fatal("could not write to blobs TOC file"); } /* @@ -706,8 +710,8 @@ _EndLOs(ArchiveHandle *AH, TocEntry *te) { lclContext *ctx = (lclContext *) AH->formatData; - if (cfclose(ctx->LOsTocFH) != 0) - pg_fatal("could not close LOs TOC file: %m"); + if (DestroyCompressFileHandle(ctx->LOsTocFH) != 0) + pg_fatal("could not close blobs TOC file: %m"); ctx->LOsTocFH = NULL; } diff --git a/src/bin/pg_dump/t/002_pg_dump.pl b/src/bin/pg_dump/t/002_pg_dump.pl index 2eeef2a4783..f3ba9263213 100644 --- a/src/bin/pg_dump/t/002_pg_dump.pl +++ b/src/bin/pg_dump/t/002_pg_dump.pl @@ -94,7 +94,7 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/compression_gzip_custom.dump", ], - expected => qr/Compression: 1/, + expected => qr/Compression: gzip/, name => 'data content is gzip-compressed' }, }, @@ -239,8 +239,8 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/defaults_custom_format.dump", ], expected => $supports_gzip ? - qr/Compression: -1/ : - qr/Compression: 0/, + qr/Compression: gzip/ : + qr/Compression: none/, name => 'data content is gzip-compressed by default if available', }, }, @@ -264,8 +264,8 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/defaults_dir_format", ], expected => $supports_gzip ? - qr/Compression: -1/ : - qr/Compression: 0/, + qr/Compression: gzip/ : + qr/Compression: none/, name => 'data content is gzip-compressed by default', }, glob_patterns => [ diff --git a/src/tools/pginclude/cpluspluscheck b/src/tools/pginclude/cpluspluscheck index fe432e2cccc..62f3e9a81d4 100755 --- a/src/tools/pginclude/cpluspluscheck +++ b/src/tools/pginclude/cpluspluscheck @@ -150,6 +150,7 @@ do # pg_dump is not C++-clean because it uses "public" and "namespace" # as field names, which is unfortunate but we won't change it now. + test "$f" = src/bin/pg_dump/compress_gzip.h && continue test "$f" = src/bin/pg_dump/compress_io.h && continue test "$f" = src/bin/pg_dump/parallel.h && continue test "$f" = src/bin/pg_dump/pg_backup_archiver.h && continue diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 23bafec5f79..840191d680b 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -428,6 +428,7 @@ CompiledExprState CompositeIOData CompositeTypeStmt CompoundAffixFlag +CompressFileHandle CompressionLocation CompressorState ComputeXidHorizonsResult @@ -1034,6 +1035,7 @@ GucStack GucStackState GucStringAssignHook GucStringCheckHook +GzipCompressorState HANDLE HASHACTION HASHBUCKET -- 2.25.1 [text/x-diff] 0004-f.patch (1.3K, ../../[email protected]/5-0004-f.patch) download | inline diff: From 100ff6665ddc4965d9cc4c0f2cd03d9b17a46099 Mon Sep 17 00:00:00 2001 From: Justin Pryzby <[email protected]> Date: Sat, 14 Jan 2023 10:31:47 -0600 Subject: [PATCH 4/7] f! --- src/bin/pg_dump/compress_gzip.c | 9 +-------- src/bin/pg_dump/pg_backup_archiver.c | 1 - 2 files changed, 1 insertion(+), 9 deletions(-) diff --git a/src/bin/pg_dump/compress_gzip.c b/src/bin/pg_dump/compress_gzip.c index 37c841c5a9b..b00be32f2e9 100644 --- a/src/bin/pg_dump/compress_gzip.c +++ b/src/bin/pg_dump/compress_gzip.c @@ -291,17 +291,10 @@ static int Gzip_close(CompressFileHandle *CFH) { gzFile gzfp = (gzFile) CFH->private_data; - int save_errno; - int ret; CFH->private_data = NULL; - ret = gzclose(gzfp); - - save_errno = errno; - errno = save_errno; - - return ret; + return gzclose(gzfp); } static int diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index 06f0b46cbfc..7f06beff61c 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -385,7 +385,6 @@ RestoreArchive(Archive *AHX) */ supports_compression = true; if (AH->compression_spec.algorithm != PG_COMPRESSION_NONE && - AH->compression_spec.algorithm == PG_COMPRESSION_GZIP && AH->PrintTocDataPtr != NULL) { for (te = AH->toc->next; te != AH->toc; te = te->next) -- 2.25.1 [text/x-diff] 0005-Add-LZ4-compression-in-pg_-dump-restore.patch (28.8K, ../../[email protected]/6-0005-Add-LZ4-compression-in-pg_-dump-restore.patch) download | inline diff: From efba6161c6d6849e5ff7cd922b1572a0e27e76b7 Mon Sep 17 00:00:00 2001 From: Georgios Kokolatos <[email protected]> Date: Wed, 21 Dec 2022 09:49:36 +0000 Subject: [PATCH 5/7] Add LZ4 compression in pg_{dump|restore} Within compress_lz4.{c,h} the streaming API and a file API compression is implemented.. The first one, is aimed at inlined use cases and thus simple lz4.h calls can be used directly. The second one is generating output, or is parsing input, which can be read/generated via the lz4 utility. Wherever the LZ4F api does not implement all the functionality corresponding to fread(), fwrite(), fgets(), fgetc(), feof(), and fclose(), it has been implemented localy. --- doc/src/sgml/ref/pg_dump.sgml | 13 +- src/bin/pg_dump/Makefile | 2 + src/bin/pg_dump/compress_io.c | 11 +- src/bin/pg_dump/compress_lz4.c | 618 +++++++++++++++++++++++++++ src/bin/pg_dump/compress_lz4.h | 22 + src/bin/pg_dump/meson.build | 8 +- src/bin/pg_dump/pg_backup_archiver.c | 14 +- src/bin/pg_dump/pg_dump.c | 5 +- src/bin/pg_dump/t/002_pg_dump.pl | 82 +++- src/tools/pginclude/cpluspluscheck | 1 + src/tools/pgindent/typedefs.list | 1 + 11 files changed, 756 insertions(+), 21 deletions(-) create mode 100644 src/bin/pg_dump/compress_lz4.c create mode 100644 src/bin/pg_dump/compress_lz4.h diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml index 2c938cd7e14..49d218905fb 100644 --- a/doc/src/sgml/ref/pg_dump.sgml +++ b/doc/src/sgml/ref/pg_dump.sgml @@ -330,9 +330,10 @@ PostgreSQL documentation machine-readable format that <application>pg_restore</application> can read. A directory format archive can be manipulated with standard Unix tools; for example, files in an uncompressed archive - can be compressed with the <application>gzip</application> tool. - This format is compressed by default and also supports parallel - dumps. + can be compressed with the <application>gzip</application> or + <application>lz4</application>tool. + This format is compressed by default using <literal>gzip</literal> + and also supports parallel dumps. </para> </listitem> </varlistentry> @@ -654,7 +655,7 @@ PostgreSQL documentation <para> Specify the compression method and/or the compression level to use. The compression method can be set to <literal>gzip</literal> or - <literal>none</literal> for no compression. + <literal>lz4</literal> or <literal>none</literal> for no compression. A compression detail string can optionally be specified. If the detail string is an integer, it specifies the compression level. Otherwise, it should be a comma-separated list of items, each of the @@ -675,8 +676,8 @@ PostgreSQL documentation individual table-data segments, and the default is to compress using <literal>gzip</literal> at a moderate level. For plain text output, setting a nonzero compression level causes the entire output file to be compressed, - as though it had been fed through <application>gzip</application>; but the default - is not to compress. + as though it had been fed through <application>gzip</application> or + <application>lz4</application>; but the default is not to compress. </para> <para> The tar archive format currently does not support compression at all. diff --git a/src/bin/pg_dump/Makefile b/src/bin/pg_dump/Makefile index 7a19f5d6172..a1401377ab9 100644 --- a/src/bin/pg_dump/Makefile +++ b/src/bin/pg_dump/Makefile @@ -17,6 +17,7 @@ top_builddir = ../../.. include $(top_builddir)/src/Makefile.global export GZIP_PROGRAM=$(GZIP) +export LZ4 export with_icu override CPPFLAGS := -I$(libpq_srcdir) $(CPPFLAGS) @@ -25,6 +26,7 @@ LDFLAGS_INTERNAL += -L$(top_builddir)/src/fe_utils -lpgfeutils $(libpq_pgport) OBJS = \ $(WIN32RES) \ compress_gzip.o \ + compress_lz4.o \ compress_io.o \ dumputils.o \ parallel.o \ diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index 576a8653193..8ebefd1ed13 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -60,6 +60,7 @@ #include "compress_io.h" #include "compress_gzip.h" +#include "compress_lz4.h" #include "pg_backup_utils.h" #ifdef HAVE_LIBZ @@ -137,7 +138,7 @@ AllocateCompressor(const pg_compress_specification *compression_spec, InitCompressorGzip(cs, compression_spec); break; case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); + InitCompressorLZ4(cs, compression_spec); break; case PG_COMPRESSION_ZSTD: pg_fatal("compression with %s is not yet supported", "ZSTD"); @@ -326,7 +327,7 @@ InitCompressFileHandle(const pg_compress_specification *compression_spec) InitCompressGzip(CFH, compression_spec); break; case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); + InitCompressLZ4(CFH, compression_spec); break; case PG_COMPRESSION_ZSTD: pg_fatal("compression with %s is not yet supported", "ZSTD"); @@ -340,12 +341,12 @@ InitCompressFileHandle(const pg_compress_specification *compression_spec) * Open a file for reading. 'path' is the file to open, and 'mode' should * be either "r" or "rb". * - * If the file at 'path' does not exist, we append the "{.gz,.lz4}" suffix (i + * If the file at 'path' does not exist, we append the ".gz" suffix (if * 'path' doesn't already have it) and try again. So if you pass "foo" as - * 'path', this will open either "foo" or "foo.gz" or "foo.lz4", trying in that - * order. + * 'path', this will open either "foo" or "foo.gz", trying in that order. * * On failure, return NULL with an error code in errno. + * */ CompressFileHandle * InitDiscoverCompressFileHandle(const char *path, const char *mode) diff --git a/src/bin/pg_dump/compress_lz4.c b/src/bin/pg_dump/compress_lz4.c new file mode 100644 index 00000000000..c97e16187a0 --- /dev/null +++ b/src/bin/pg_dump/compress_lz4.c @@ -0,0 +1,618 @@ +/*------------------------------------------------------------------------- + * + * compress_lz4.c + * Routines for archivers to write an uncompressed or compressed data + * stream. + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_lz4.c + * + *------------------------------------------------------------------------- + */ +#include "postgres_fe.h" +#include "pg_backup_utils.h" + +#include "compress_lz4.h" + +#ifdef USE_LZ4 +#include <lz4.h> +#include <lz4frame.h> + +#define LZ4_OUT_SIZE (4 * 1024) +#define LZ4_IN_SIZE (16 * 1024) + +/* + * LZ4F_HEADER_SIZE_MAX first appeared in v1.7.5 of the library. + * Redefine it for installations with a lesser version. + */ +#ifndef LZ4F_HEADER_SIZE_MAX +#define LZ4F_HEADER_SIZE_MAX 32 +#endif + +/*---------------------- + * Compressor API + *---------------------- + */ + +typedef struct LZ4CompressorState +{ + char *outbuf; + size_t outsize; +} LZ4CompressorState; + +/* Private routines that support LZ4 compressed data I/O */ +static void ReadDataFromArchiveLZ4(ArchiveHandle *AH, CompressorState *cs); +static void WriteDataToArchiveLZ4(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen); +static void EndCompressorLZ4(ArchiveHandle *AH, CompressorState *cs); + +static void +ReadDataFromArchiveLZ4(ArchiveHandle *AH, CompressorState *cs) +{ + LZ4_streamDecode_t lz4StreamDecode; + char *buf; + char *decbuf; + size_t buflen; + size_t cnt; + + buflen = LZ4_IN_SIZE; + buf = pg_malloc(buflen); + decbuf = pg_malloc(buflen); + + LZ4_setStreamDecode(&lz4StreamDecode, NULL, 0); + + while ((cnt = cs->readF(AH, &buf, &buflen))) + { + int decBytes = LZ4_decompress_safe_continue(&lz4StreamDecode, + buf, decbuf, + cnt, buflen); + + ahwrite(decbuf, 1, decBytes, AH); + } + + pg_free(buf); + pg_free(decbuf); +} + +static void +WriteDataToArchiveLZ4(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen) +{ + LZ4CompressorState *LZ4cs = (LZ4CompressorState *) cs->private_data; + size_t compressed; + size_t requiredsize = LZ4_compressBound(dLen); + + if (requiredsize > LZ4cs->outsize) + { + LZ4cs->outbuf = pg_realloc(LZ4cs->outbuf, requiredsize); + LZ4cs->outsize = requiredsize; + } + + compressed = LZ4_compress_default(data, LZ4cs->outbuf, + dLen, LZ4cs->outsize); + + if (compressed <= 0) + pg_fatal("failed to LZ4 compress data"); + + cs->writeF(AH, LZ4cs->outbuf, compressed); +} + +static void +EndCompressorLZ4(ArchiveHandle *AH, CompressorState *cs) +{ + LZ4CompressorState *LZ4cs; + + LZ4cs = (LZ4CompressorState *) cs->private_data; + if (LZ4cs) + { + pg_free(LZ4cs->outbuf); + pg_free(LZ4cs); + cs->private_data = NULL; + } +} + + +/* Public routines that support LZ4 compressed data I/O */ +void +InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec) +{ + cs->readData = ReadDataFromArchiveLZ4; + cs->writeData = WriteDataToArchiveLZ4; + cs->end = EndCompressorLZ4; + + cs->compression_spec = compression_spec; + + /* Will be lazy init'd */ + cs->private_data = pg_malloc0(sizeof(LZ4CompressorState)); +} + +/*---------------------- + * Compress File API + *---------------------- + */ + +/* + * State needed for LZ4 (de)compression using the CompressFileHandle API. + */ +typedef struct LZ4File +{ + FILE *fp; + + LZ4F_preferences_t prefs; + + LZ4F_compressionContext_t ctx; + LZ4F_decompressionContext_t dtx; + + bool inited; + bool compressing; + + size_t buflen; + char *buffer; + + size_t overflowalloclen; + size_t overflowlen; + char *overflowbuf; + + size_t errcode; +} LZ4File; + +/* + * LZ4 equivalent to feof() or gzeof(). The end of file + * is reached if there is no decompressed output in the + * overflow buffer and the end of the file is reached. + */ +static int +LZ4File_eof(CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + + return fs->overflowlen == 0 && feof(fs->fp); +} + +static const char * +LZ4File_get_error(CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + const char *errmsg; + + if (LZ4F_isError(fs->errcode)) + errmsg = LZ4F_getErrorName(fs->errcode); + else + errmsg = strerror(errno); + + return errmsg; +} + +/* + * Prepare an already alloc'ed LZ4File struct for subsequent calls. + * + * It creates the nessary contexts for the operations. When compressing, + * it additionally writes the LZ4 header in the output stream. + */ +static int +LZ4File_init(LZ4File * fs, int size, bool compressing) +{ + size_t status; + + if (fs->inited) + return 0; + + fs->compressing = compressing; + fs->inited = true; + + if (fs->compressing) + { + fs->buflen = LZ4F_compressBound(LZ4_IN_SIZE, &fs->prefs); + if (fs->buflen < LZ4F_HEADER_SIZE_MAX) + fs->buflen = LZ4F_HEADER_SIZE_MAX; + + status = LZ4F_createCompressionContext(&fs->ctx, LZ4F_VERSION); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return 1; + } + + fs->buffer = pg_malloc(fs->buflen); + status = LZ4F_compressBegin(fs->ctx, fs->buffer, fs->buflen, + &fs->prefs); + + if (LZ4F_isError(status)) + { + fs->errcode = status; + return 1; + } + + if (fwrite(fs->buffer, 1, status, fs->fp) != status) + { + errno = errno ? : ENOSPC; + return 1; + } + } + else + { + status = LZ4F_createDecompressionContext(&fs->dtx, LZ4F_VERSION); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return 1; + } + + fs->buflen = size > LZ4_OUT_SIZE ? size : LZ4_OUT_SIZE; + fs->buffer = pg_malloc(fs->buflen); + + fs->overflowalloclen = fs->buflen; + fs->overflowbuf = pg_malloc(fs->overflowalloclen); + fs->overflowlen = 0; + } + + return 0; +} + +/* + * Read already decompressed content from the overflow buffer into 'ptr' up to + * 'size' bytes, if available. If the eol_flag is set, then stop at the first + * occurance of the new line char prior to 'size' bytes. + * + * Any unread content in the overflow buffer, is moved to the beginning. + */ +static int +LZ4File_read_overflow(LZ4File * fs, void *ptr, int size, bool eol_flag) +{ + char *p; + int readlen = 0; + + if (fs->overflowlen == 0) + return 0; + + if (fs->overflowlen >= size) + readlen = size; + else + readlen = fs->overflowlen; + + if (eol_flag && (p = memchr(fs->overflowbuf, '\n', readlen))) + /* Include the line terminating char */ + readlen = p - fs->overflowbuf + 1; + + memcpy(ptr, fs->overflowbuf, readlen); + fs->overflowlen -= readlen; + + if (fs->overflowlen > 0) + memmove(fs->overflowbuf, fs->overflowbuf + readlen, fs->overflowlen); + + return readlen; +} + +/* + * The workhorse for reading decompressed content out of an LZ4 compressed + * stream. + * + * It will read up to 'ptrsize' decompressed content, or up to the new line char + * if found first when the eol_flag is set. It is possible that the decompressed + * output generated by reading any compressed input via the LZ4F API, exceeds + * 'ptrsize'. Any exceeding decompressed content is stored at an overflow + * buffer within LZ4File. Of course, when the function is called, it will first + * try to consume any decompressed content already present in the overflow + * buffer, before decompressing new content. + */ +static int +LZ4File_read_internal(LZ4File * fs, void *ptr, int ptrsize, bool eol_flag) +{ + size_t dsize = 0; + size_t rsize; + size_t size = ptrsize; + bool eol_found = false; + + void *readbuf; + + /* Lazy init */ + if (!fs->inited && LZ4File_init(fs, size, false /* decompressing */ )) + return -1; + + /* Verfiy that there is enough space in the outbuf */ + if (size > fs->buflen) + { + fs->buflen = size; + fs->buffer = pg_realloc(fs->buffer, size); + } + + /* use already decompressed content if available */ + dsize = LZ4File_read_overflow(fs, ptr, size, eol_flag); + if (dsize == size || (eol_flag && memchr(ptr, '\n', dsize))) + return dsize; + + readbuf = pg_malloc(size); + + do + { + char *rp; + char *rend; + + rsize = fread(readbuf, 1, size, fs->fp); + if (rsize < size && !feof(fs->fp)) + return -1; + + rp = (char *) readbuf; + rend = (char *) readbuf + rsize; + + while (rp < rend) + { + size_t status; + size_t outlen = fs->buflen; + size_t read_remain = rend - rp; + + memset(fs->buffer, 0, outlen); + status = LZ4F_decompress(fs->dtx, fs->buffer, &outlen, + rp, &read_remain, NULL); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return -1; + } + + rp += read_remain; + + /* + * fill in what space is available in ptr if the eol flag is set, + * either skip if one already found or fill up to EOL if present + * in the outbuf + */ + if (outlen > 0 && dsize < size && eol_found == false) + { + char *p; + size_t lib = (eol_flag == 0) ? size - dsize : size - 1 - dsize; + size_t len = outlen < lib ? outlen : lib; + + if (eol_flag == true && + (p = memchr(fs->buffer, '\n', outlen)) && + (size_t) (p - fs->buffer + 1) <= len) + { + len = p - fs->buffer + 1; + eol_found = true; + } + + memcpy((char *) ptr + dsize, fs->buffer, len); + dsize += len; + + /* move what did not fit, if any, at the begining of the buf */ + if (len < outlen) + memmove(fs->buffer, fs->buffer + len, outlen - len); + outlen -= len; + } + + /* if there is available output, save it */ + if (outlen > 0) + { + while (fs->overflowlen + outlen > fs->overflowalloclen) + { + fs->overflowalloclen *= 2; + fs->overflowbuf = pg_realloc(fs->overflowbuf, + fs->overflowalloclen); + } + + memcpy(fs->overflowbuf + fs->overflowlen, fs->buffer, outlen); + fs->overflowlen += outlen; + } + } + } while (rsize == size && dsize < size && eol_found == 0); + + pg_free(readbuf); + + return (int) dsize; +} + +/* + * Compress size bytes from ptr and write them to the stream. + */ +static size_t +LZ4File_write(const void *ptr, size_t size, CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + size_t status; + int remaining = size; + + if (!fs->inited && LZ4File_init(fs, size, true)) + return -1; + + while (remaining > 0) + { + int chunk = remaining < LZ4_IN_SIZE ? remaining : LZ4_IN_SIZE; + + remaining -= chunk; + + status = LZ4F_compressUpdate(fs->ctx, fs->buffer, fs->buflen, + ptr, chunk, NULL); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return -1; + } + + if (fwrite(fs->buffer, 1, status, fs->fp) != status) + { + errno = errno ? : ENOSPC; + return 1; + } + } + + return size; +} + +/* + * fread() equivalent implementation for LZ4 compressed files. + */ +static size_t +LZ4File_read(void *ptr, size_t size, CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + int ret; + + ret = LZ4File_read_internal(fs, ptr, size, false); + if (ret != size && !LZ4File_eof(CFH)) + pg_fatal("could not read from input file: %s", LZ4File_get_error(CFH)); + + return ret; +} + +/* + * fgetc() equivalent implementation for LZ4 compressed files. + */ +static int +LZ4File_getc(CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + unsigned char c; + + if (LZ4File_read_internal(fs, &c, 1, false) != 1) + { + if (!LZ4File_eof(CFH)) + pg_fatal("could not read from input file: %s", LZ4File_get_error(CFH)); + else + pg_fatal("could not read from input file: end of file"); + } + + return c; +} + +/* + * fgets() equivalent implementation for LZ4 compressed files. + */ +static char * +LZ4File_gets(char *ptr, int size, CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + size_t dsize; + + dsize = LZ4File_read_internal(fs, ptr, size, true); + if (dsize < 0) + pg_fatal("could not read from input file: %s", LZ4File_get_error(CFH)); + + /* Done reading */ + if (dsize == 0) + return NULL; + + return ptr; +} + +/* + * Finalize (de)compression of a stream. When compressing it will write any + * remaining content and/or generated footer from the LZ4 API. + */ +static int +LZ4File_close(CompressFileHandle *CFH) +{ + FILE *fp; + LZ4File *fs = (LZ4File *) CFH->private_data; + size_t status; + int ret; + + fp = fs->fp; + if (fs->inited) + { + if (fs->compressing) + { + status = LZ4F_compressEnd(fs->ctx, fs->buffer, fs->buflen, NULL); + if (LZ4F_isError(status)) + pg_fatal("failed to end compression: %s", + LZ4F_getErrorName(status)); + else if ((ret = fwrite(fs->buffer, 1, status, fs->fp)) != status) + { + errno = errno ? : ENOSPC; + WRITE_ERROR_EXIT; + } + + status = LZ4F_freeCompressionContext(fs->ctx); + if (LZ4F_isError(status)) + pg_fatal("failed to end compression: %s", + LZ4F_getErrorName(status)); + } + else + { + status = LZ4F_freeDecompressionContext(fs->dtx); + if (LZ4F_isError(status)) + pg_fatal("failed to end decompression: %s", + LZ4F_getErrorName(status)); + pg_free(fs->overflowbuf); + } + + pg_free(fs->buffer); + } + + pg_free(fs); + + return fclose(fp); +} + +static int +LZ4File_open(const char *path, int fd, const char *mode, + CompressFileHandle *CFH) +{ + FILE *fp; + LZ4File *lz4fp = (LZ4File *) CFH->private_data; + + if (fd >= 0) + fp = fdopen(fd, mode); + else + fp = fopen(path, mode); + if (fp == NULL) + { + lz4fp->errcode = errno; + return 1; + } + + lz4fp->fp = fp; + + return 0; +} + +static int +LZ4File_open_write(const char *path, const char *mode, CompressFileHandle *CFH) +{ + char *fname; + int ret; + + fname = psprintf("%s.lz4", path); + ret = CFH->open_func(fname, -1, mode, CFH); + pg_free(fname); + + return ret; +} + +void +InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +{ + LZ4File *lz4fp; + + CFH->open_func = LZ4File_open; + CFH->open_write_func = LZ4File_open_write; + CFH->read_func = LZ4File_read; + CFH->write_func = LZ4File_write; + CFH->gets_func = LZ4File_gets; + CFH->getc_func = LZ4File_getc; + CFH->eof_func = LZ4File_eof; + CFH->close_func = LZ4File_close; + CFH->get_error_func = LZ4File_get_error; + + CFH->compression_spec = compression_spec; + lz4fp = pg_malloc0(sizeof(*lz4fp)); + if (CFH->compression_spec.level >= 0) + lz4fp->prefs.compressionLevel = CFH->compression_spec.level; + + CFH->private_data = lz4fp; +} +#else /* USE_LZ4 */ +void +InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec) +{ + pg_fatal("this build does not support compression with %s", "LZ4"); +} + +void +InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +{ + pg_fatal("this build does not support compression with %s", "LZ4"); +} +#endif /* USE_LZ4 */ diff --git a/src/bin/pg_dump/compress_lz4.h b/src/bin/pg_dump/compress_lz4.h new file mode 100644 index 00000000000..74595db1b98 --- /dev/null +++ b/src/bin/pg_dump/compress_lz4.h @@ -0,0 +1,22 @@ +/*------------------------------------------------------------------------- + * + * compress_lz4.h + * Interface to compress_io.c routines + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_lz4.h + * + *------------------------------------------------------------------------- + */ +#ifndef _COMPRESS_LZ4_H_ +#define _COMPRESS_LZ4_H_ + +#include "compress_io.h" + +extern void InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec); +extern void InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec); + +#endif /* _COMPRESS_LZ4_H_ */ diff --git a/src/bin/pg_dump/meson.build b/src/bin/pg_dump/meson.build index aa2c91829c0..473d40d456f 100644 --- a/src/bin/pg_dump/meson.build +++ b/src/bin/pg_dump/meson.build @@ -3,6 +3,7 @@ pg_dump_common_sources = files( 'compress_io.c', 'compress_gzip.c', + 'compress_lz4.c', 'dumputils.c', 'parallel.c', 'pg_backup_archiver.c', @@ -17,7 +18,7 @@ pg_dump_common_sources = files( pg_dump_common = static_library('libpgdump_common', pg_dump_common_sources, c_pch: pch_postgres_fe_h, - dependencies: [frontend_code, libpq, zlib], + dependencies: [frontend_code, libpq, lz4, zlib], kwargs: internal_lib_args, ) @@ -85,7 +86,10 @@ tests += { 'sd': meson.current_source_dir(), 'bd': meson.current_build_dir(), 'tap': { - 'env': {'GZIP_PROGRAM': gzip.path()}, + 'env': { + 'GZIP_PROGRAM': gzip.path(), + 'LZ4': program_lz4.found() ? program_lz4.path() : '', + }, 'tests': [ 't/001_basic.pl', 't/002_pg_dump.pl', diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index 7f06beff61c..2d406a5f0e3 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -394,6 +394,10 @@ RestoreArchive(Archive *AHX) #ifndef HAVE_LIBZ if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) supports_compression = false; +#endif +#ifndef USE_LZ4 + if (AH->compression_spec.algorithm == PG_COMPRESSION_LZ4) + supports_compression = false; #endif if (supports_compression == false) pg_fatal("cannot restore from compressed archive (compression not supported in this installation)"); @@ -2073,7 +2077,7 @@ _discoverArchiveFormat(ArchiveHandle *AH) /* * Check if the specified archive is a directory. If so, check if - * there's a "toc.dat" (or "toc.dat.gz") file in it. + * there's a "toc.dat" (or "toc.dat.{gz,lz4}") file in it. */ if (stat(AH->fSpec, &st) == 0 && S_ISDIR(st.st_mode)) { @@ -2083,6 +2087,10 @@ _discoverArchiveFormat(ArchiveHandle *AH) #ifdef HAVE_LIBZ if (_fileExistsInDirectory(AH->fSpec, "toc.dat.gz")) return AH->format; +#endif +#ifdef USE_LZ4 + if (_fileExistsInDirectory(AH->fSpec, "toc.dat.lz4")) + return AH->format; #endif pg_fatal("directory \"%s\" does not appear to be a valid archive (\"toc.dat\" does not exist)", AH->fSpec); @@ -3746,6 +3754,10 @@ ReadHead(ArchiveHandle *AH) #ifndef HAVE_LIBZ if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) unsupported = true; +#endif +#ifndef USE_LZ4 + if (AH->compression_spec.algorithm == PG_COMPRESSION_LZ4) + unsupported = true; #endif if (unsupported) pg_fatal("archive is compressed, but this installation does not support compression"); diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c index 20f73729fac..224d2c900ce 100644 --- a/src/bin/pg_dump/pg_dump.c +++ b/src/bin/pg_dump/pg_dump.c @@ -715,13 +715,12 @@ main(int argc, char **argv) case PG_COMPRESSION_NONE: /* fallthrough */ case PG_COMPRESSION_GZIP: + /* fallthrough */ + case PG_COMPRESSION_LZ4: break; case PG_COMPRESSION_ZSTD: pg_fatal("compression with %s is not yet supported", "ZSTD"); break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; } /* diff --git a/src/bin/pg_dump/t/002_pg_dump.pl b/src/bin/pg_dump/t/002_pg_dump.pl index f3ba9263213..f497ec60407 100644 --- a/src/bin/pg_dump/t/002_pg_dump.pl +++ b/src/bin/pg_dump/t/002_pg_dump.pl @@ -139,6 +139,80 @@ my %pgdump_runs = ( args => [ '-d', "$tempdir/compression_gzip_plain.sql.gz", ], }, }, + + # Do not use --no-sync to give test coverage for data sync. + compression_lz4_custom => { + test_key => 'compression', + compile_option => 'lz4', + dump_cmd => [ + 'pg_dump', '--format=custom', + '--compress=lz4', "--file=$tempdir/compression_lz4_custom.dump", + 'postgres', + ], + restore_cmd => [ + 'pg_restore', + "--file=$tempdir/compression_lz4_custom.sql", + "$tempdir/compression_lz4_custom.dump", + ], + command_like => { + command => [ + 'pg_restore', + '-l', "$tempdir/compression_lz4_custom.dump", + ], + expected => qr/Compression: lz4/, + name => 'data content is lz4 compressed' + }, + }, + + # Do not use --no-sync to give test coverage for data sync. + compression_lz4_dir => { + test_key => 'compression', + compile_option => 'lz4', + dump_cmd => [ + 'pg_dump', '--jobs=2', + '--format=directory', '--compress=lz4:1', + "--file=$tempdir/compression_lz4_dir", 'postgres', + ], + # Give coverage for manually compressed blob.toc files during + # restore. + compress_cmd => { + program => $ENV{'LZ4'}, + args => [ + '-z', '-f', '--rm', + "$tempdir/compression_lz4_dir/blobs.toc", + "$tempdir/compression_lz4_dir/blobs.toc.lz4", + ], + }, + # Verify that data files where compressed + glob_patterns => [ + "$tempdir/compression_lz4_dir/toc.dat", + "$tempdir/compression_lz4_dir/*.dat.lz4", + ], + restore_cmd => [ + 'pg_restore', '--jobs=2', + "--file=$tempdir/compression_lz4_dir.sql", + "$tempdir/compression_lz4_dir", + ], + }, + + compression_lz4_plain => { + test_key => 'compression', + compile_option => 'lz4', + dump_cmd => [ + 'pg_dump', '--format=plain', '--compress=lz4', + "--file=$tempdir/compression_lz4_plain.sql.lz4", 'postgres', + ], + # Decompress the generated file to run through the tests. + compress_cmd => { + program => $ENV{'LZ4'}, + args => [ + '-d', '-f', + "$tempdir/compression_lz4_plain.sql.lz4", + "$tempdir/compression_lz4_plain.sql", + ], + }, + }, + clean => { dump_cmd => [ 'pg_dump', @@ -4175,11 +4249,11 @@ foreach my $run (sort keys %pgdump_runs) my $run_db = 'postgres'; # Skip command-level tests for gzip if there is no support for it. - if ( defined($pgdump_runs{$run}->{compile_option}) - && $pgdump_runs{$run}->{compile_option} eq 'gzip' - && !$supports_gzip) + if ($pgdump_runs{$run}->{compile_option} && + ($pgdump_runs{$run}->{compile_option} eq 'gzip' && !$supports_gzip) || + ($pgdump_runs{$run}->{compile_option} eq 'lz4' && !$supports_lz4)) { - note "$run: skipped due to no gzip support"; + note "$run: skipped due to no $pgdump_runs{$run}->{compile_option} support"; next; } diff --git a/src/tools/pginclude/cpluspluscheck b/src/tools/pginclude/cpluspluscheck index 62f3e9a81d4..2b461e797c6 100755 --- a/src/tools/pginclude/cpluspluscheck +++ b/src/tools/pginclude/cpluspluscheck @@ -152,6 +152,7 @@ do # as field names, which is unfortunate but we won't change it now. test "$f" = src/bin/pg_dump/compress_gzip.h && continue test "$f" = src/bin/pg_dump/compress_io.h && continue + test "$f" = src/bin/pg_dump/compress_lz4.h && continue test "$f" = src/bin/pg_dump/parallel.h && continue test "$f" = src/bin/pg_dump/pg_backup_archiver.h && continue test "$f" = src/bin/pg_dump/pg_dump.h && continue diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 840191d680b..232228d427c 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -1381,6 +1381,7 @@ LWLock LWLockHandle LWLockMode LWLockPadded +LZ4CompressorState LZ4F_compressionContext_t LZ4F_decompressOptions_t LZ4F_decompressionContext_t -- 2.25.1 [text/x-diff] 0006-f.patch (6.8K, ../../[email protected]/7-0006-f.patch) download | inline diff: From cf1efb67c49c5e31d77049f5469967dd750ee8c9 Mon Sep 17 00:00:00 2001 From: Justin Pryzby <[email protected]> Date: Sat, 14 Jan 2023 10:45:02 -0600 Subject: [PATCH 6/7] f! --- src/bin/pg_dump/compress_lz4.c | 34 ++++++++++++++++---------------- src/bin/pg_dump/compress_lz4.h | 4 ++-- src/bin/pg_dump/t/002_pg_dump.pl | 2 +- 3 files changed, 20 insertions(+), 20 deletions(-) diff --git a/src/bin/pg_dump/compress_lz4.c b/src/bin/pg_dump/compress_lz4.c index c97e16187a0..0e259a6251a 100644 --- a/src/bin/pg_dump/compress_lz4.c +++ b/src/bin/pg_dump/compress_lz4.c @@ -117,13 +117,13 @@ EndCompressorLZ4(ArchiveHandle *AH, CompressorState *cs) /* Public routines that support LZ4 compressed data I/O */ void -InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec) +InitCompressorLZ4(CompressorState *cs, const pg_compress_specification *compression_spec) { cs->readData = ReadDataFromArchiveLZ4; cs->writeData = WriteDataToArchiveLZ4; cs->end = EndCompressorLZ4; - cs->compression_spec = compression_spec; + cs->compression_spec = *compression_spec; /* Will be lazy init'd */ cs->private_data = pg_malloc0(sizeof(LZ4CompressorState)); @@ -189,7 +189,7 @@ LZ4File_get_error(CompressFileHandle *CFH) /* * Prepare an already alloc'ed LZ4File struct for subsequent calls. * - * It creates the nessary contexts for the operations. When compressing, + * It creates the necessary contexts for the operations. When compressing, * it additionally writes the LZ4 header in the output stream. */ static int @@ -228,7 +228,7 @@ LZ4File_init(LZ4File * fs, int size, bool compressing) if (fwrite(fs->buffer, 1, status, fs->fp) != status) { - errno = errno ? : ENOSPC; + errno = errno ? errno : ENOSPC; return 1; } } @@ -255,7 +255,7 @@ LZ4File_init(LZ4File * fs, int size, bool compressing) /* * Read already decompressed content from the overflow buffer into 'ptr' up to * 'size' bytes, if available. If the eol_flag is set, then stop at the first - * occurance of the new line char prior to 'size' bytes. + * occurrence of the new line char prior to 'size' bytes. * * Any unread content in the overflow buffer, is moved to the beginning. */ @@ -309,10 +309,10 @@ LZ4File_read_internal(LZ4File * fs, void *ptr, int ptrsize, bool eol_flag) void *readbuf; /* Lazy init */ - if (!fs->inited && LZ4File_init(fs, size, false /* decompressing */ )) + if (LZ4File_init(fs, size, false /* decompressing */ )) return -1; - /* Verfiy that there is enough space in the outbuf */ + /* Verify that there is enough space in the outbuf */ if (size > fs->buflen) { fs->buflen = size; @@ -363,10 +363,10 @@ LZ4File_read_internal(LZ4File * fs, void *ptr, int ptrsize, bool eol_flag) if (outlen > 0 && dsize < size && eol_found == false) { char *p; - size_t lib = (eol_flag == 0) ? size - dsize : size - 1 - dsize; + size_t lib = eol_flag ? size - 1 - dsize : size - dsize ; size_t len = outlen < lib ? outlen : lib; - if (eol_flag == true && + if (eol_flag && (p = memchr(fs->buffer, '\n', outlen)) && (size_t) (p - fs->buffer + 1) <= len) { @@ -377,7 +377,7 @@ LZ4File_read_internal(LZ4File * fs, void *ptr, int ptrsize, bool eol_flag) memcpy((char *) ptr + dsize, fs->buffer, len); dsize += len; - /* move what did not fit, if any, at the begining of the buf */ + /* move what did not fit, if any, at the beginning of the buf */ if (len < outlen) memmove(fs->buffer, fs->buffer + len, outlen - len); outlen -= len; @@ -414,7 +414,7 @@ LZ4File_write(const void *ptr, size_t size, CompressFileHandle *CFH) size_t status; int remaining = size; - if (!fs->inited && LZ4File_init(fs, size, true)) + if (LZ4File_init(fs, size, true)) return -1; while (remaining > 0) @@ -433,7 +433,7 @@ LZ4File_write(const void *ptr, size_t size, CompressFileHandle *CFH) if (fwrite(fs->buffer, 1, status, fs->fp) != status) { - errno = errno ? : ENOSPC; + errno = errno ? errno : ENOSPC; return 1; } } @@ -520,7 +520,7 @@ LZ4File_close(CompressFileHandle *CFH) LZ4F_getErrorName(status)); else if ((ret = fwrite(fs->buffer, 1, status, fs->fp)) != status) { - errno = errno ? : ENOSPC; + errno = errno ? errno : ENOSPC; WRITE_ERROR_EXIT; } @@ -582,7 +582,7 @@ LZ4File_open_write(const char *path, const char *mode, CompressFileHandle *CFH) } void -InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification *compression_spec) { LZ4File *lz4fp; @@ -596,7 +596,7 @@ InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compres CFH->close_func = LZ4File_close; CFH->get_error_func = LZ4File_get_error; - CFH->compression_spec = compression_spec; + CFH->compression_spec = *compression_spec; lz4fp = pg_malloc0(sizeof(*lz4fp)); if (CFH->compression_spec.level >= 0) lz4fp->prefs.compressionLevel = CFH->compression_spec.level; @@ -605,13 +605,13 @@ InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compres } #else /* USE_LZ4 */ void -InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec) +InitCompressorLZ4(CompressorState *cs, const pg_compress_specification *compression_spec) { pg_fatal("this build does not support compression with %s", "LZ4"); } void -InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification *compression_spec) { pg_fatal("this build does not support compression with %s", "LZ4"); } diff --git a/src/bin/pg_dump/compress_lz4.h b/src/bin/pg_dump/compress_lz4.h index 74595db1b98..69a3d9c171f 100644 --- a/src/bin/pg_dump/compress_lz4.h +++ b/src/bin/pg_dump/compress_lz4.h @@ -16,7 +16,7 @@ #include "compress_io.h" -extern void InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec); -extern void InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec); +extern void InitCompressorLZ4(CompressorState *cs, const pg_compress_specification *compression_spec); +extern void InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification *compression_spec); #endif /* _COMPRESS_LZ4_H_ */ diff --git a/src/bin/pg_dump/t/002_pg_dump.pl b/src/bin/pg_dump/t/002_pg_dump.pl index f497ec60407..263995a2b7a 100644 --- a/src/bin/pg_dump/t/002_pg_dump.pl +++ b/src/bin/pg_dump/t/002_pg_dump.pl @@ -183,7 +183,7 @@ my %pgdump_runs = ( "$tempdir/compression_lz4_dir/blobs.toc.lz4", ], }, - # Verify that data files where compressed + # Verify that data files were compressed glob_patterns => [ "$tempdir/compression_lz4_dir/toc.dat", "$tempdir/compression_lz4_dir/*.dat.lz4", -- 2.25.1 [text/x-diff] 0007-TMP-pg_dump-use-lz4-by-default-for-CI-only.patch (2.0K, ../../[email protected]/8-0007-TMP-pg_dump-use-lz4-by-default-for-CI-only.patch) download | inline diff: From d2fe3c9c5bb1fdce5c58af2773c603358085a0bb Mon Sep 17 00:00:00 2001 From: Justin Pryzby <[email protected]> Date: Wed, 4 Jan 2023 21:21:53 -0600 Subject: [PATCH 7/7] TMP: pg_dump: use lz4 by default, for CI only --- src/bin/pg_dump/pg_dump.c | 7 +++++-- src/bin/pg_dump/t/002_pg_dump.pl | 8 ++++---- 2 files changed, 9 insertions(+), 6 deletions(-) diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c index 224d2c900ce..cf5083c432f 100644 --- a/src/bin/pg_dump/pg_dump.c +++ b/src/bin/pg_dump/pg_dump.c @@ -733,8 +733,11 @@ main(int argc, char **argv) #ifdef HAVE_LIBZ parse_compress_specification(PG_COMPRESSION_GZIP, NULL, &compression_spec); -#else - /* Nothing to do in the default case */ +#endif + +#ifdef USE_LZ4 + parse_compress_specification(PG_COMPRESSION_LZ4, NULL, + &compression_spec); #endif } diff --git a/src/bin/pg_dump/t/002_pg_dump.pl b/src/bin/pg_dump/t/002_pg_dump.pl index 263995a2b7a..3485ebca57d 100644 --- a/src/bin/pg_dump/t/002_pg_dump.pl +++ b/src/bin/pg_dump/t/002_pg_dump.pl @@ -313,9 +313,9 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/defaults_custom_format.dump", ], expected => $supports_gzip ? - qr/Compression: gzip/ : + qr/Compression: lz4/ : qr/Compression: none/, - name => 'data content is gzip-compressed by default if available', + name => 'data content is lz4-compressed by default if available', }, }, @@ -338,7 +338,7 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/defaults_dir_format", ], expected => $supports_gzip ? - qr/Compression: gzip/ : + qr/Compression: lz4/ : qr/Compression: none/, name => 'data content is gzip-compressed by default', }, @@ -346,7 +346,7 @@ my %pgdump_runs = ( "$tempdir/defaults_dir_format/toc.dat", "$tempdir/defaults_dir_format/blobs.toc", $supports_gzip ? - "$tempdir/defaults_dir_format/*.dat.gz" : + "$tempdir/defaults_dir_format/*.dat.lz4" : "$tempdir/defaults_dir_format/*.dat", ], }, -- 2.25.1 ^ permalink raw reply [nested|flat] 29+ messages in thread
* Re: Add LZ4 compression in pg_dump @ 2023-01-15 22:32 Justin Pryzby <[email protected]> parent: Justin Pryzby <[email protected]> 1 sibling, 0 replies; 29+ messages in thread From: Justin Pryzby @ 2023-01-15 22:32 UTC (permalink / raw) To: [email protected]; +Cc: Michael Paquier <[email protected]>; [email protected]; Rachel Heaton <[email protected]> On Sat, Jan 14, 2023 at 03:43:08PM -0600, Justin Pryzby wrote: > On Sun, Jan 08, 2023 at 01:45:25PM -0600, Justin Pryzby wrote: > > pg_compress_specification is being passed by value, but I think it > > should be passed as a pointer, as is done everywhere else. > > ISTM that was an issue with 5e73a6048, affecting a few public and > private functions. I wrote a pre-preparatory patch which changes to > pass by reference. I updated 001 to change SetOutput() to pass by reference, too (before, that ended up in the 002 patch). I can't see any issue in 002 other than the == GZIP change (the fix for which I'd previously included in a later patch). > One more thing: some comments still refer to the cfopen API, which this > patch removes. > > > There were "LZ4" comments and file extension stuff in the preparatory > > commit. But now it seems like you *removed* them in the LZ4 commit > > (where it actually belongs) rather than *moving* it from the > > prior/parent commit *to* the lz4 commit. I recommend to run something > > like "git diff @{1}" whenever doing this kind of patch surgery. > > TODO I addressed that in the fixup commits 005 and 007. -- Justin Attachments: [text/x-diff] 0001-pg_dump-pass-pg_compress_specification-as-a-pointer.patch (13.8K, ../../[email protected]/2-0001-pg_dump-pass-pg_compress_specification-as-a-pointer.patch) download | inline diff: From b822e6ace433235232a74cbe50af514271f9e49d Mon Sep 17 00:00:00 2001 From: Justin Pryzby <[email protected]> Date: Sat, 14 Jan 2023 10:58:23 -0600 Subject: [PATCH 1/8] pg_dump: pass pg_compress_specification as a pointer.. ..as is done everywhere else. --- src/bin/pg_dump/compress_io.c | 30 +++++++++++++-------------- src/bin/pg_dump/compress_io.h | 8 +++---- src/bin/pg_dump/pg_backup.h | 2 +- src/bin/pg_dump/pg_backup_archiver.c | 24 ++++++++++----------- src/bin/pg_dump/pg_backup_custom.c | 6 +++--- src/bin/pg_dump/pg_backup_directory.c | 8 +++---- src/bin/pg_dump/pg_dump.c | 2 +- 7 files changed, 40 insertions(+), 40 deletions(-) diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index 7a2c80bbc4c..e5107c75874 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -94,19 +94,19 @@ static void WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, /* Allocate a new compressor */ CompressorState * -AllocateCompressor(const pg_compress_specification compression_spec, +AllocateCompressor(const pg_compress_specification *compression_spec, WriteFunc writeF) { CompressorState *cs; #ifndef HAVE_LIBZ - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) + if (compression_spec->algorithm == PG_COMPRESSION_GZIP) pg_fatal("this build does not support compression with %s", "gzip"); #endif cs = (CompressorState *) pg_malloc0(sizeof(CompressorState)); cs->writeF = writeF; - cs->compression_spec = compression_spec; + cs->compression_spec = *compression_spec; /* * Perform compression algorithm specific initialization. @@ -125,12 +125,12 @@ AllocateCompressor(const pg_compress_specification compression_spec, */ void ReadDataFromArchive(ArchiveHandle *AH, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, ReadFunc readF) { - if (compression_spec.algorithm == PG_COMPRESSION_NONE) + if (compression_spec->algorithm == PG_COMPRESSION_NONE) ReadDataFromArchiveNone(AH, readF); - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) + if (compression_spec->algorithm == PG_COMPRESSION_GZIP) { #ifdef HAVE_LIBZ ReadDataFromArchiveZlib(AH, readF); @@ -432,13 +432,13 @@ cfopen_read(const char *path, const char *mode) if (hasSuffix(path, ".gz")) { compression_spec.algorithm = PG_COMPRESSION_GZIP; - fp = cfopen(path, mode, compression_spec); + fp = cfopen(path, mode, &compression_spec); } else #endif { compression_spec.algorithm = PG_COMPRESSION_NONE; - fp = cfopen(path, mode, compression_spec); + fp = cfopen(path, mode, &compression_spec); #ifdef HAVE_LIBZ if (fp == NULL) { @@ -446,7 +446,7 @@ cfopen_read(const char *path, const char *mode) fname = psprintf("%s.gz", path); compression_spec.algorithm = PG_COMPRESSION_GZIP; - fp = cfopen(fname, mode, compression_spec); + fp = cfopen(fname, mode, &compression_spec); free_keep_errno(fname); } #endif @@ -467,11 +467,11 @@ cfopen_read(const char *path, const char *mode) */ cfp * cfopen_write(const char *path, const char *mode, - const pg_compress_specification compression_spec) + const pg_compress_specification *compression_spec) { cfp *fp; - if (compression_spec.algorithm == PG_COMPRESSION_NONE) + if (compression_spec->algorithm == PG_COMPRESSION_NONE) fp = cfopen(path, mode, compression_spec); else { @@ -497,20 +497,20 @@ cfopen_write(const char *path, const char *mode, */ cfp * cfopen(const char *path, const char *mode, - const pg_compress_specification compression_spec) + const pg_compress_specification *compression_spec) { cfp *fp = pg_malloc(sizeof(cfp)); - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) + if (compression_spec->algorithm == PG_COMPRESSION_GZIP) { #ifdef HAVE_LIBZ - if (compression_spec.level != Z_DEFAULT_COMPRESSION) + if (compression_spec->level != Z_DEFAULT_COMPRESSION) { /* user has specified a compression level, so tell zlib to use it */ char mode_compression[32]; snprintf(mode_compression, sizeof(mode_compression), "%s%d", - mode, compression_spec.level); + mode, compression_spec->level); fp->compressedfp = gzopen(path, mode_compression); } else diff --git a/src/bin/pg_dump/compress_io.h b/src/bin/pg_dump/compress_io.h index a429dc4789d..34f4e5e1e14 100644 --- a/src/bin/pg_dump/compress_io.h +++ b/src/bin/pg_dump/compress_io.h @@ -40,10 +40,10 @@ typedef size_t (*ReadFunc) (ArchiveHandle *AH, char **buf, size_t *buflen); /* struct definition appears in compress_io.c */ typedef struct CompressorState CompressorState; -extern CompressorState *AllocateCompressor(const pg_compress_specification compression_spec, +extern CompressorState *AllocateCompressor(const pg_compress_specification *compression_spec, WriteFunc writeF); extern void ReadDataFromArchive(ArchiveHandle *AH, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, ReadFunc readF); extern void WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, const void *data, size_t dLen); @@ -53,10 +53,10 @@ extern void EndCompressor(ArchiveHandle *AH, CompressorState *cs); typedef struct cfp cfp; extern cfp *cfopen(const char *path, const char *mode, - const pg_compress_specification compression_spec); + const pg_compress_specification *compression_spec); extern cfp *cfopen_read(const char *path, const char *mode); extern cfp *cfopen_write(const char *path, const char *mode, - const pg_compress_specification compression_spec); + const pg_compress_specification *compression_spec); extern int cfread(void *ptr, int size, cfp *fp); extern int cfwrite(const void *ptr, int size, cfp *fp); extern int cfgetc(cfp *fp); diff --git a/src/bin/pg_dump/pg_backup.h b/src/bin/pg_dump/pg_backup.h index aba780ef4b1..216e24e7ec5 100644 --- a/src/bin/pg_dump/pg_backup.h +++ b/src/bin/pg_dump/pg_backup.h @@ -305,7 +305,7 @@ extern Archive *OpenArchive(const char *FileSpec, const ArchiveFormat fmt); /* Create a new archive */ extern Archive *CreateArchive(const char *FileSpec, const ArchiveFormat fmt, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, bool dosync, ArchiveMode mode, SetupWorkerPtrType setupDumpWorker); diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index 7f7a0f1ce7b..b82bad107f8 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -70,7 +70,7 @@ typedef struct _parallelReadyList static ArchiveHandle *_allocAH(const char *FileSpec, const ArchiveFormat fmt, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, bool dosync, ArchiveMode mode, SetupWorkerPtrType setupWorkerPtr); static void _getObjectDescription(PQExpBuffer buf, const TocEntry *te); @@ -100,7 +100,7 @@ static int RestoringToDB(ArchiveHandle *AH); static void dump_lo_buf(ArchiveHandle *AH); static void dumpTimestamp(ArchiveHandle *AH, const char *msg, time_t tim); static void SetOutput(ArchiveHandle *AH, const char *filename, - const pg_compress_specification compression_spec); + const pg_compress_specification *compression_spec); static OutputContext SaveOutput(ArchiveHandle *AH); static void RestoreOutput(ArchiveHandle *AH, OutputContext savedContext); @@ -241,7 +241,7 @@ setupRestoreWorker(Archive *AHX) /* Public */ Archive * CreateArchive(const char *FileSpec, const ArchiveFormat fmt, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, bool dosync, ArchiveMode mode, SetupWorkerPtrType setupDumpWorker) @@ -261,7 +261,7 @@ OpenArchive(const char *FileSpec, const ArchiveFormat fmt) pg_compress_specification compression_spec = {0}; compression_spec.algorithm = PG_COMPRESSION_NONE; - AH = _allocAH(FileSpec, fmt, compression_spec, true, + AH = _allocAH(FileSpec, fmt, &compression_spec, true, archModeRead, setupRestoreWorker); return (Archive *) AH; @@ -469,7 +469,7 @@ RestoreArchive(Archive *AHX) */ sav = SaveOutput(AH); if (ropt->filename || ropt->compression_spec.algorithm != PG_COMPRESSION_NONE) - SetOutput(AH, ropt->filename, ropt->compression_spec); + SetOutput(AH, ropt->filename, &ropt->compression_spec); ahprintf(AH, "--\n-- PostgreSQL database dump\n--\n\n"); @@ -1137,7 +1137,7 @@ PrintTOCSummary(Archive *AHX) sav = SaveOutput(AH); if (ropt->filename) - SetOutput(AH, ropt->filename, out_compression_spec); + SetOutput(AH, ropt->filename, &out_compression_spec); if (strftime(stamp_str, sizeof(stamp_str), PGDUMP_STRFTIME_FMT, localtime(&AH->createDate)) == 0) @@ -1501,7 +1501,7 @@ archprintf(Archive *AH, const char *fmt,...) static void SetOutput(ArchiveHandle *AH, const char *filename, - const pg_compress_specification compression_spec) + const pg_compress_specification *compression_spec) { int fn; @@ -1524,12 +1524,12 @@ SetOutput(ArchiveHandle *AH, const char *filename, /* If compression explicitly requested, use gzopen */ #ifdef HAVE_LIBZ - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) + if (compression_spec->algorithm == PG_COMPRESSION_GZIP) { char fmode[14]; /* Don't use PG_BINARY_x since this is zlib */ - sprintf(fmode, "wb%d", compression_spec.level); + sprintf(fmode, "wb%d", compression_spec->level); if (fn >= 0) AH->OF = gzdopen(dup(fn), fmode); else @@ -2214,7 +2214,7 @@ _discoverArchiveFormat(ArchiveHandle *AH) */ static ArchiveHandle * _allocAH(const char *FileSpec, const ArchiveFormat fmt, - const pg_compress_specification compression_spec, + const pg_compress_specification *compression_spec, bool dosync, ArchiveMode mode, SetupWorkerPtrType setupWorkerPtr) { @@ -2266,7 +2266,7 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, AH->toc->prev = AH->toc; AH->mode = mode; - AH->compression_spec = compression_spec; + AH->compression_spec = *compression_spec; AH->dosync = dosync; memset(&(AH->sqlparse), 0, sizeof(AH->sqlparse)); @@ -2281,7 +2281,7 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, * Force stdin/stdout into binary mode if that is what we are using. */ #ifdef WIN32 - if ((fmt != archNull || compression_spec.algorithm != PG_COMPRESSION_NONE) && + if ((fmt != archNull || compression_spec->algorithm != PG_COMPRESSION_NONE) && (AH->fSpec == NULL || strcmp(AH->fSpec, "") == 0)) { if (mode == archModeWrite) diff --git a/src/bin/pg_dump/pg_backup_custom.c b/src/bin/pg_dump/pg_backup_custom.c index d1e54644a94..0e87444de85 100644 --- a/src/bin/pg_dump/pg_backup_custom.c +++ b/src/bin/pg_dump/pg_backup_custom.c @@ -298,7 +298,7 @@ _StartData(ArchiveHandle *AH, TocEntry *te) _WriteByte(AH, BLK_DATA); /* Block type */ WriteInt(AH, te->dumpId); /* For sanity check */ - ctx->cs = AllocateCompressor(AH->compression_spec, _CustomWriteFunc); + ctx->cs = AllocateCompressor(&AH->compression_spec, _CustomWriteFunc); } /* @@ -377,7 +377,7 @@ _StartLO(ArchiveHandle *AH, TocEntry *te, Oid oid) WriteInt(AH, oid); - ctx->cs = AllocateCompressor(AH->compression_spec, _CustomWriteFunc); + ctx->cs = AllocateCompressor(&AH->compression_spec, _CustomWriteFunc); } /* @@ -566,7 +566,7 @@ _PrintTocData(ArchiveHandle *AH, TocEntry *te) static void _PrintData(ArchiveHandle *AH) { - ReadDataFromArchive(AH, AH->compression_spec, _CustomReadFunc); + ReadDataFromArchive(AH, &AH->compression_spec, _CustomReadFunc); } static void diff --git a/src/bin/pg_dump/pg_backup_directory.c b/src/bin/pg_dump/pg_backup_directory.c index 6800c3cceef..ffb8a0e4dd7 100644 --- a/src/bin/pg_dump/pg_backup_directory.c +++ b/src/bin/pg_dump/pg_backup_directory.c @@ -328,7 +328,7 @@ _StartData(ArchiveHandle *AH, TocEntry *te) setFilePath(AH, fname, tctx->filename); ctx->dataFH = cfopen_write(fname, PG_BINARY_W, - AH->compression_spec); + &AH->compression_spec); if (ctx->dataFH == NULL) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -584,7 +584,7 @@ _CloseArchive(ArchiveHandle *AH) /* The TOC is always created uncompressed */ compression_spec.algorithm = PG_COMPRESSION_NONE; - tocFH = cfopen_write(fname, PG_BINARY_W, compression_spec); + tocFH = cfopen_write(fname, PG_BINARY_W, &compression_spec); if (tocFH == NULL) pg_fatal("could not open output file \"%s\": %m", fname); ctx->dataFH = tocFH; @@ -649,7 +649,7 @@ _StartLOs(ArchiveHandle *AH, TocEntry *te) /* The LO TOC file is never compressed */ compression_spec.algorithm = PG_COMPRESSION_NONE; - ctx->LOsTocFH = cfopen_write(fname, "ab", compression_spec); + ctx->LOsTocFH = cfopen_write(fname, "ab", &compression_spec); if (ctx->LOsTocFH == NULL) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -667,7 +667,7 @@ _StartLO(ArchiveHandle *AH, TocEntry *te, Oid oid) snprintf(fname, MAXPGPATH, "%s/blob_%u.dat", ctx->directory, oid); - ctx->dataFH = cfopen_write(fname, PG_BINARY_W, AH->compression_spec); + ctx->dataFH = cfopen_write(fname, PG_BINARY_W, &AH->compression_spec); if (ctx->dataFH == NULL) pg_fatal("could not open output file \"%s\": %m", fname); diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c index 2c0a9699729..20f73729fac 100644 --- a/src/bin/pg_dump/pg_dump.c +++ b/src/bin/pg_dump/pg_dump.c @@ -751,7 +751,7 @@ main(int argc, char **argv) pg_fatal("parallel backup only supported by the directory format"); /* Open the output file */ - fout = CreateArchive(filename, archiveFormat, compression_spec, + fout = CreateArchive(filename, archiveFormat, &compression_spec, dosync, archiveMode, setupDumpWorker); /* Make dump options accessible right away */ -- 2.25.1 [text/x-diff] 0002-Prepare-pg_dump-internals-for-additional-compression.patch (22.6K, ../../[email protected]/3-0002-Prepare-pg_dump-internals-for-additional-compression.patch) download | inline diff: From 7f14216f4c46017f48efc73cb4e3021b573c5391 Mon Sep 17 00:00:00 2001 From: Georgios Kokolatos <[email protected]> Date: Wed, 21 Dec 2022 09:49:23 +0000 Subject: [PATCH 2/8] Prepare pg_dump internals for additional compression methods. Commit bf9aa490db introduced cfp in compress_io.{c,h} with the intent of unifying compression related code and allow for the introduction of additional archive formats. However, pg_backup_archiver.c was not using that API. This commit teaches pg_backup_archiver.c about it and is using it throughout. --- src/bin/pg_dump/compress_io.c | 389 +++++++++++++++++++-------- src/bin/pg_dump/compress_io.h | 2 + src/bin/pg_dump/pg_backup_archiver.c | 128 +++------ src/bin/pg_dump/pg_backup_archiver.h | 27 +- 4 files changed, 320 insertions(+), 226 deletions(-) diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index e5107c75874..8d0bec08d7c 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -56,6 +56,10 @@ #include "compress_io.h" #include "pg_backup_utils.h" +#ifdef HAVE_LIBZ +#include <zlib.h> +#endif + /*---------------------- * Compressor API *---------------------- @@ -128,15 +132,24 @@ ReadDataFromArchive(ArchiveHandle *AH, const pg_compress_specification *compression_spec, ReadFunc readF) { - if (compression_spec->algorithm == PG_COMPRESSION_NONE) - ReadDataFromArchiveNone(AH, readF); - if (compression_spec->algorithm == PG_COMPRESSION_GZIP) + switch (compression_spec->algorithm) { + case PG_COMPRESSION_NONE: + ReadDataFromArchiveNone(AH, readF); + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - ReadDataFromArchiveZlib(AH, readF); + ReadDataFromArchiveZlib(AH, readF); #else - pg_fatal("this build does not support compression with %s", "gzip"); + pg_fatal("this build does not support compression with %s", "gzip"); #endif + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } } @@ -149,6 +162,9 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, { switch (cs->compression_spec.algorithm) { + case PG_COMPRESSION_NONE: + WriteDataToArchiveNone(AH, cs, data, dLen); + break; case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ WriteDataToArchiveZlib(AH, cs, data, dLen); @@ -156,13 +172,11 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, pg_fatal("this build does not support compression with %s", "gzip"); #endif break; - case PG_COMPRESSION_NONE: - WriteDataToArchiveNone(AH, cs, data, dLen); - break; case PG_COMPRESSION_LZ4: - /* fallthrough */ + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; case PG_COMPRESSION_ZSTD: - pg_fatal("invalid compression method"); + pg_fatal("compression with %s is not yet supported", "ZSTD"); break; } } @@ -173,10 +187,26 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, void EndCompressor(ArchiveHandle *AH, CompressorState *cs) { + switch (cs->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (cs->compression_spec.algorithm == PG_COMPRESSION_GZIP) - EndCompressorZlib(AH, cs); + EndCompressorZlib(AH, cs); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + free(cs); } @@ -391,10 +421,8 @@ WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, */ struct cfp { - FILE *uncompressedfp; -#ifdef HAVE_LIBZ - gzFile compressedfp; -#endif + pg_compress_specification compression_spec; + void *fp; }; #ifdef HAVE_LIBZ @@ -490,127 +518,202 @@ cfopen_write(const char *path, const char *mode, } /* - * Opens file 'path' in 'mode'. If compression is GZIP, the file - * is opened with libz gzopen(), otherwise with plain fopen(). + * This is the workhorse for cfopen() or cfdopen(). It opens file 'path' or + * associates a stream 'fd', if 'fd' is a valid descriptor, in 'mode'. The + * descriptor is not dup'ed and it is the caller's responsibility to do so. + * The caller must verify that the 'compress_algorithm' is supported by the + * current build. * * On failure, return NULL with an error code in errno. */ -cfp * -cfopen(const char *path, const char *mode, - const pg_compress_specification *compression_spec) +static cfp * +cfopen_internal(const char *path, int fd, const char *mode, + const pg_compress_specification *compression_spec) { cfp *fp = pg_malloc(sizeof(cfp)); - if (compression_spec->algorithm == PG_COMPRESSION_GZIP) + fp->compression_spec = *compression_spec; + + switch (compression_spec->algorithm) { -#ifdef HAVE_LIBZ - if (compression_spec->level != Z_DEFAULT_COMPRESSION) - { - /* user has specified a compression level, so tell zlib to use it */ - char mode_compression[32]; + case PG_COMPRESSION_NONE: + if (fd >= 0) + fp->fp = fdopen(fd, mode); + else + fp->fp = fopen(path, mode); + if (fp->fp == NULL) + { + free_keep_errno(fp); + fp = NULL; + } - snprintf(mode_compression, sizeof(mode_compression), "%s%d", - mode, compression_spec->level); - fp->compressedfp = gzopen(path, mode_compression); - } - else - { - /* don't specify a level, just use the zlib default */ - fp->compressedfp = gzopen(path, mode); - } + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + if (compression_spec->level != Z_DEFAULT_COMPRESSION) + { + /* + * user has specified a compression level, so tell zlib to use + * it + */ + char mode_compression[32]; + + snprintf(mode_compression, sizeof(mode_compression), "%s%d", + mode, compression_spec->level); + if (fd >= 0) + fp->fp = gzdopen(fd, mode_compression); + else + fp->fp = gzopen(path, mode_compression); + } + else + { + /* don't specify a level, just use the zlib default */ + if (fd >= 0) + fp->fp = gzdopen(fd, mode); + else + fp->fp = gzopen(path, mode); + } - fp->uncompressedfp = NULL; - if (fp->compressedfp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } + if (fp->fp == NULL) + { + free_keep_errno(fp); + fp = NULL; + } #else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - } - else - { -#ifdef HAVE_LIBZ - fp->compressedfp = NULL; + pg_fatal("this build does not support compression with %s", "gzip"); #endif - fp->uncompressedfp = fopen(path, mode); - if (fp->uncompressedfp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } return fp; } +cfp * +cfopen(const char *path, const char *mode, + const pg_compress_specification *compression_spec) +{ + return cfopen_internal(path, -1, mode, compression_spec); +} + +cfp * +cfdopen(int fd, const char *mode, + const pg_compress_specification *compression_spec) +{ + return cfopen_internal(NULL, fd, mode, compression_spec); +} int cfread(void *ptr, int size, cfp *fp) { - int ret; + int ret = 0; if (size == 0) return 0; -#ifdef HAVE_LIBZ - if (fp->compressedfp) + switch (fp->compression_spec.algorithm) { - ret = gzread(fp->compressedfp, ptr, size); - if (ret != size && !gzeof(fp->compressedfp)) - { - int errnum; - const char *errmsg = gzerror(fp->compressedfp, &errnum); + case PG_COMPRESSION_NONE: + ret = fread(ptr, 1, size, (FILE *) fp->fp); + if (ret != size && !feof((FILE *) fp->fp)) + READ_ERROR_EXIT((FILE *) fp->fp); - pg_fatal("could not read from input file: %s", - errnum == Z_ERRNO ? strerror(errno) : errmsg); - } - } - else + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + ret = gzread((gzFile) fp->fp, ptr, size); + if (ret != size && !gzeof((gzFile) fp->fp)) + { + int errnum; + const char *errmsg = gzerror((gzFile) fp->fp, &errnum); + + pg_fatal("could not read from input file: %s", + errnum == Z_ERRNO ? strerror(errno) : errmsg); + } +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - { - ret = fread(ptr, 1, size, fp->uncompressedfp); - if (ret != size && !feof(fp->uncompressedfp)) - READ_ERROR_EXIT(fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } + return ret; } int cfwrite(const void *ptr, int size, cfp *fp) { + int ret = 0; + + switch (fp->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + ret = fwrite(ptr, 1, size, (FILE *) fp->fp); + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (fp->compressedfp) - return gzwrite(fp->compressedfp, ptr, size); - else + ret = gzwrite((gzFile) fp->fp, ptr, size); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - return fwrite(ptr, 1, size, fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + + return ret; } int cfgetc(cfp *fp) { - int ret; + int ret = 0; -#ifdef HAVE_LIBZ - if (fp->compressedfp) + switch (fp->compression_spec.algorithm) { - ret = gzgetc(fp->compressedfp); - if (ret == EOF) - { - if (!gzeof(fp->compressedfp)) - pg_fatal("could not read from input file: %s", strerror(errno)); - else - pg_fatal("could not read from input file: end of file"); - } - } - else + case PG_COMPRESSION_NONE: + ret = fgetc((FILE *) fp->fp); + if (ret == EOF) + READ_ERROR_EXIT((FILE *) fp->fp); + + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + ret = gzgetc((gzFile) fp->fp); + if (ret == EOF) + { + if (!gzeof((gzFile) fp->fp)) + pg_fatal("could not read from input file: %s", strerror(errno)); + else + pg_fatal("could not read from input file: end of file"); + } +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - { - ret = fgetc(fp->uncompressedfp); - if (ret == EOF) - READ_ERROR_EXIT(fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } return ret; @@ -619,65 +722,119 @@ cfgetc(cfp *fp) char * cfgets(cfp *fp, char *buf, int len) { + char *ret = NULL; + + switch (fp->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + ret = fgets(buf, len, (FILE *) fp->fp); + + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (fp->compressedfp) - return gzgets(fp->compressedfp, buf, len); - else + ret = gzgets((gzFile) fp->fp, buf, len); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - return fgets(buf, len, fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + + return ret; } int cfclose(cfp *fp) { - int result; + int ret = 0; if (fp == NULL) { errno = EBADF; return EOF; } -#ifdef HAVE_LIBZ - if (fp->compressedfp) + + switch (fp->compression_spec.algorithm) { - result = gzclose(fp->compressedfp); - fp->compressedfp = NULL; - } - else + case PG_COMPRESSION_NONE: + ret = fclose((FILE *) fp->fp); + fp->fp = NULL; + + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + ret = gzclose((gzFile) fp->fp); + fp->fp = NULL; +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - { - result = fclose(fp->uncompressedfp); - fp->uncompressedfp = NULL; + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } + free_keep_errno(fp); - return result; + return ret; } int cfeof(cfp *fp) { + int ret = 0; + + switch (fp->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + ret = feof((FILE *) fp->fp); + + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (fp->compressedfp) - return gzeof(fp->compressedfp); - else + ret = gzeof((gzFile) fp->fp); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - return feof(fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + + return ret; } const char * get_cfp_error(cfp *fp) { -#ifdef HAVE_LIBZ - if (fp->compressedfp) + if (fp->compression_spec.algorithm == PG_COMPRESSION_GZIP) { +#ifdef HAVE_LIBZ int errnum; - const char *errmsg = gzerror(fp->compressedfp, &errnum); + const char *errmsg = gzerror((gzFile) fp->fp, &errnum); if (errnum != Z_ERRNO) return errmsg; - } +#else + pg_fatal("this build does not support compression with %s", "gzip"); #endif + } + return strerror(errno); } diff --git a/src/bin/pg_dump/compress_io.h b/src/bin/pg_dump/compress_io.h index 34f4e5e1e14..768096c820d 100644 --- a/src/bin/pg_dump/compress_io.h +++ b/src/bin/pg_dump/compress_io.h @@ -54,6 +54,8 @@ typedef struct cfp cfp; extern cfp *cfopen(const char *path, const char *mode, const pg_compress_specification *compression_spec); +extern cfp *cfdopen(int fd, const char *mode, + const pg_compress_specification *compression_spec); extern cfp *cfopen_read(const char *path, const char *mode); extern cfp *cfopen_write(const char *path, const char *mode, const pg_compress_specification *compression_spec); diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index b82bad107f8..5164b57f042 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -31,6 +31,7 @@ #endif #include "common/string.h" +#include "compress_io.h" #include "dumputils.h" #include "fe_utils/string_utils.h" #include "lib/stringinfo.h" @@ -43,13 +44,6 @@ #define TEXT_DUMP_HEADER "--\n-- PostgreSQL database dump\n--\n\n" #define TEXT_DUMPALL_HEADER "--\n-- PostgreSQL database cluster dump\n--\n\n" -/* state needed to save/restore an archive's output target */ -typedef struct _outputContext -{ - void *OF; - int gzOut; -} OutputContext; - /* * State for tracking TocEntrys that are ready to process during a parallel * restore. (This used to be a list, and we still call it that, though now @@ -101,8 +95,8 @@ static void dump_lo_buf(ArchiveHandle *AH); static void dumpTimestamp(ArchiveHandle *AH, const char *msg, time_t tim); static void SetOutput(ArchiveHandle *AH, const char *filename, const pg_compress_specification *compression_spec); -static OutputContext SaveOutput(ArchiveHandle *AH); -static void RestoreOutput(ArchiveHandle *AH, OutputContext savedContext); +static cfp *SaveOutput(ArchiveHandle *AH); +static void RestoreOutput(ArchiveHandle *AH, cfp *savedOutput); static int restore_toc_entry(ArchiveHandle *AH, TocEntry *te, bool is_parallel); static void restore_toc_entries_prefork(ArchiveHandle *AH, @@ -277,11 +271,8 @@ CloseArchive(Archive *AHX) AH->ClosePtr(AH); /* Close the output */ - errno = 0; /* in case gzclose() doesn't set it */ - if (AH->gzOut) - res = GZCLOSE(AH->OF); - else if (AH->OF != stdout) - res = fclose(AH->OF); + errno = 0; + res = cfclose(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); @@ -362,8 +353,9 @@ RestoreArchive(Archive *AHX) ArchiveHandle *AH = (ArchiveHandle *) AHX; RestoreOptions *ropt = AH->public.ropt; bool parallel_mode; + bool supports_compression; TocEntry *te; - OutputContext sav; + cfp *sav; AH->stage = STAGE_INITIALIZING; @@ -391,17 +383,24 @@ RestoreArchive(Archive *AHX) /* * Make sure we won't need (de)compression we haven't got */ -#ifndef HAVE_LIBZ - if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP && + supports_compression = true; + if (AH->compression_spec.algorithm != PG_COMPRESSION_NONE && + AH->compression_spec.algorithm == PG_COMPRESSION_GZIP && AH->PrintTocDataPtr != NULL) { for (te = AH->toc->next; te != AH->toc; te = te->next) { if (te->hadDumper && (te->reqs & REQ_DATA) != 0) - pg_fatal("cannot restore from compressed archive (compression not supported in this installation)"); + { +#ifndef HAVE_LIBZ + if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) + supports_compression = false; +#endif + if (supports_compression == false) + pg_fatal("cannot restore from compressed archive (compression not supported in this installation)"); + } } } -#endif /* * Prepare index arrays, so we can assume we have them throughout restore. @@ -1128,7 +1127,7 @@ PrintTOCSummary(Archive *AHX) TocEntry *te; pg_compress_specification out_compression_spec = {0}; teSection curSection; - OutputContext sav; + cfp *sav; const char *fmtName; char stamp_str[64]; @@ -1503,58 +1502,32 @@ static void SetOutput(ArchiveHandle *AH, const char *filename, const pg_compress_specification *compression_spec) { - int fn; + const char *mode; + int fn = -1; if (filename) { if (strcmp(filename, "-") == 0) fn = fileno(stdout); - else - fn = -1; } else if (AH->FH) fn = fileno(AH->FH); else if (AH->fSpec) { - fn = -1; filename = AH->fSpec; } else fn = fileno(stdout); - /* If compression explicitly requested, use gzopen */ -#ifdef HAVE_LIBZ - if (compression_spec->algorithm == PG_COMPRESSION_GZIP) - { - char fmode[14]; + if (AH->mode == archModeAppend) + mode = PG_BINARY_A; + else + mode = PG_BINARY_W; - /* Don't use PG_BINARY_x since this is zlib */ - sprintf(fmode, "wb%d", compression_spec->level); - if (fn >= 0) - AH->OF = gzdopen(dup(fn), fmode); - else - AH->OF = gzopen(filename, fmode); - AH->gzOut = 1; - } + if (fn >= 0) + AH->OF = cfdopen(dup(fn), mode, compression_spec); else -#endif - { /* Use fopen */ - if (AH->mode == archModeAppend) - { - if (fn >= 0) - AH->OF = fdopen(dup(fn), PG_BINARY_A); - else - AH->OF = fopen(filename, PG_BINARY_A); - } - else - { - if (fn >= 0) - AH->OF = fdopen(dup(fn), PG_BINARY_W); - else - AH->OF = fopen(filename, PG_BINARY_W); - } - AH->gzOut = 0; - } + AH->OF = cfopen(filename, mode, compression_spec); if (!AH->OF) { @@ -1565,33 +1538,24 @@ SetOutput(ArchiveHandle *AH, const char *filename, } } -static OutputContext +static cfp * SaveOutput(ArchiveHandle *AH) { - OutputContext sav; - - sav.OF = AH->OF; - sav.gzOut = AH->gzOut; - - return sav; + return (cfp *) AH->OF; } static void -RestoreOutput(ArchiveHandle *AH, OutputContext savedContext) +RestoreOutput(ArchiveHandle *AH, cfp *savedOutput) { int res; - errno = 0; /* in case gzclose() doesn't set it */ - if (AH->gzOut) - res = GZCLOSE(AH->OF); - else - res = fclose(AH->OF); + errno = 0; + res = cfclose(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); - AH->gzOut = savedContext.gzOut; - AH->OF = savedContext.OF; + AH->OF = savedOutput; } @@ -1715,22 +1679,17 @@ ahwrite(const void *ptr, size_t size, size_t nmemb, ArchiveHandle *AH) bytes_written = size * nmemb; } - else if (AH->gzOut) - bytes_written = GZWRITE(ptr, size, nmemb, AH->OF); else if (AH->CustomOutPtr) bytes_written = AH->CustomOutPtr(AH, ptr, size * nmemb); + /* + * If we're doing a restore, and it's direct to DB, and we're connected + * then send it to the DB. + */ + else if (RestoringToDB(AH)) + bytes_written = ExecuteSqlCommandBuf(&AH->public, (const char *) ptr, size * nmemb); else - { - /* - * If we're doing a restore, and it's direct to DB, and we're - * connected then send it to the DB. - */ - if (RestoringToDB(AH)) - bytes_written = ExecuteSqlCommandBuf(&AH->public, (const char *) ptr, size * nmemb); - else - bytes_written = fwrite(ptr, size, nmemb, AH->OF) * size; - } + bytes_written = cfwrite(ptr, size * nmemb, AH->OF); if (bytes_written != size * nmemb) WRITE_ERROR_EXIT; @@ -2219,6 +2178,7 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, SetupWorkerPtrType setupWorkerPtr) { ArchiveHandle *AH; + pg_compress_specification out_compress_spec = {0}; pg_log_debug("allocating AH for %s, format %d", FileSpec ? FileSpec : "(stdio)", fmt); @@ -2272,8 +2232,8 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, memset(&(AH->sqlparse), 0, sizeof(AH->sqlparse)); /* Open stdout with no compression for AH output handle */ - AH->gzOut = 0; - AH->OF = stdout; + out_compress_spec.algorithm = PG_COMPRESSION_NONE; + AH->OF = cfdopen(dup(fileno(stdout)), PG_BINARY_A, &out_compress_spec); /* * On Windows, we need to use binary mode to read/write non-text files, diff --git a/src/bin/pg_dump/pg_backup_archiver.h b/src/bin/pg_dump/pg_backup_archiver.h index f72446ed5b4..4725e49747b 100644 --- a/src/bin/pg_dump/pg_backup_archiver.h +++ b/src/bin/pg_dump/pg_backup_archiver.h @@ -32,30 +32,6 @@ #define LOBBUFSIZE 16384 -#ifdef HAVE_LIBZ -#include <zlib.h> -#define GZCLOSE(fh) gzclose(fh) -#define GZWRITE(p, s, n, fh) gzwrite(fh, p, (n) * (s)) -#define GZREAD(p, s, n, fh) gzread(fh, p, (n) * (s)) -#define GZEOF(fh) gzeof(fh) -#else -#define GZCLOSE(fh) fclose(fh) -#define GZWRITE(p, s, n, fh) (fwrite(p, s, n, fh) * (s)) -#define GZREAD(p, s, n, fh) fread(p, s, n, fh) -#define GZEOF(fh) feof(fh) -/* this is just the redefinition of a libz constant */ -#define Z_DEFAULT_COMPRESSION (-1) - -typedef struct _z_stream -{ - void *next_in; - void *next_out; - size_t avail_in; - size_t avail_out; -} z_stream; -typedef z_stream *z_streamp; -#endif - /* Data block types */ #define BLK_DATA 1 #define BLK_BLOBS 3 @@ -319,8 +295,7 @@ struct _archiveHandle char *fSpec; /* Archive File Spec */ FILE *FH; /* General purpose file handle */ - void *OF; - int gzOut; /* Output file */ + void *OF; /* Output file */ struct _tocEntry *toc; /* Header of circular list of TOC entries */ int tocCount; /* Number of TOC entries */ -- 2.25.1 [text/x-diff] 0003-f.patch (786B, ../../[email protected]/4-0003-f.patch) download | inline diff: From 5105ac46180d287f8df61a58bd8042f5f6aed1e0 Mon Sep 17 00:00:00 2001 From: Justin Pryzby <[email protected]> Date: Sun, 15 Jan 2023 16:27:28 -0600 Subject: [PATCH 3/8] f! --- src/bin/pg_dump/pg_backup_archiver.c | 1 - 1 file changed, 1 deletion(-) diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index 5164b57f042..d4f1e09fce6 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -385,7 +385,6 @@ RestoreArchive(Archive *AHX) */ supports_compression = true; if (AH->compression_spec.algorithm != PG_COMPRESSION_NONE && - AH->compression_spec.algorithm == PG_COMPRESSION_GZIP && AH->PrintTocDataPtr != NULL) { for (te = AH->toc->next; te != AH->toc; te = te->next) -- 2.25.1 [text/x-diff] 0004-Introduce-Compressor-API-in-pg_dump.patch (65.4K, ../../[email protected]/5-0004-Introduce-Compressor-API-in-pg_dump.patch) download | inline diff: From a9db15f09c8cbf71603bddc7d0b411e6c91411c8 Mon Sep 17 00:00:00 2001 From: Georgios Kokolatos <[email protected]> Date: Wed, 21 Dec 2022 09:49:31 +0000 Subject: [PATCH 4/8] Introduce Compressor API in pg_dump The purpose of this API is to allow for easier addition of new compression methods. CompressFileHandle is substituting the cfp* family of functions under a struct of function pointers for opening, writing, etc. The implementor of a new compression method is now able to "simply" just add those definitions. Custom compressed archives now need to store the compression algorithm in their header. This requires a bump in the version number. The level of compression is no longer stored in the dump as it is irrelevant. --- src/bin/pg_dump/Makefile | 1 + src/bin/pg_dump/compress_gzip.c | 405 ++++++++++++ src/bin/pg_dump/compress_gzip.h | 22 + src/bin/pg_dump/compress_io.c | 914 +++++++------------------- src/bin/pg_dump/compress_io.h | 71 +- src/bin/pg_dump/meson.build | 1 + src/bin/pg_dump/pg_backup_archiver.c | 102 +-- src/bin/pg_dump/pg_backup_archiver.h | 5 +- src/bin/pg_dump/pg_backup_custom.c | 23 +- src/bin/pg_dump/pg_backup_directory.c | 94 +-- src/bin/pg_dump/t/002_pg_dump.pl | 10 +- src/tools/pginclude/cpluspluscheck | 1 + src/tools/pgindent/typedefs.list | 2 + 13 files changed, 852 insertions(+), 799 deletions(-) create mode 100644 src/bin/pg_dump/compress_gzip.c create mode 100644 src/bin/pg_dump/compress_gzip.h diff --git a/src/bin/pg_dump/Makefile b/src/bin/pg_dump/Makefile index ef1ed0f3e51..7a19f5d6172 100644 --- a/src/bin/pg_dump/Makefile +++ b/src/bin/pg_dump/Makefile @@ -24,6 +24,7 @@ LDFLAGS_INTERNAL += -L$(top_builddir)/src/fe_utils -lpgfeutils $(libpq_pgport) OBJS = \ $(WIN32RES) \ + compress_gzip.o \ compress_io.o \ dumputils.o \ parallel.o \ diff --git a/src/bin/pg_dump/compress_gzip.c b/src/bin/pg_dump/compress_gzip.c new file mode 100644 index 00000000000..37c841c5a9b --- /dev/null +++ b/src/bin/pg_dump/compress_gzip.c @@ -0,0 +1,405 @@ +/*------------------------------------------------------------------------- + * + * compress_gzip.c + * Routines for archivers to write an uncompressed or compressed data + * stream. + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_gzip.c + * + *------------------------------------------------------------------------- + */ +#include "postgres_fe.h" +#include <unistd.h> + +#include "compress_gzip.h" +#include "pg_backup_utils.h" + +#ifdef HAVE_LIBZ +#include "zlib.h" + +/*---------------------- + * Compressor API + *---------------------- + */ +typedef struct GzipCompressorState +{ + z_streamp zp; + + void *outbuf; + size_t outsize; +} GzipCompressorState; + +/* Private routines that support gzip compressed data I/O */ +static void +DeflateCompressorGzip(ArchiveHandle *AH, CompressorState *cs, bool flush) +{ + GzipCompressorState *gzipcs = (GzipCompressorState *) cs->private_data; + z_streamp zp = gzipcs->zp; + void *out = gzipcs->outbuf; + int res = Z_OK; + + while (gzipcs->zp->avail_in != 0 || flush) + { + res = deflate(zp, flush ? Z_FINISH : Z_NO_FLUSH); + if (res == Z_STREAM_ERROR) + pg_fatal("could not compress data: %s", zp->msg); + if ((flush && (zp->avail_out < gzipcs->outsize)) + || (zp->avail_out == 0) + || (zp->avail_in != 0) + ) + { + /* + * Extra paranoia: avoid zero-length chunks, since a zero length + * chunk is the EOF marker in the custom format. This should never + * happen but... + */ + if (zp->avail_out < gzipcs->outsize) + { + /* + * Any write function should do its own error checking but to + * make sure we do a check here as well... + */ + size_t len = gzipcs->outsize - zp->avail_out; + + cs->writeF(AH, (char *) out, len); + } + zp->next_out = out; + zp->avail_out = gzipcs->outsize; + } + + if (res == Z_STREAM_END) + break; + } +} + +static void +EndCompressorGzip(ArchiveHandle *AH, CompressorState *cs) +{ + GzipCompressorState *gzipcs = (GzipCompressorState *) cs->private_data; + z_streamp zp; + + if (gzipcs->zp) + { + zp = gzipcs->zp; + zp->next_in = NULL; + zp->avail_in = 0; + + /* Flush any remaining data from zlib buffer */ + DeflateCompressorGzip(AH, cs, true); + + if (deflateEnd(zp) != Z_OK) + pg_fatal("could not close compression stream: %s", zp->msg); + + pg_free(gzipcs->outbuf); + pg_free(gzipcs->zp); + } + + pg_free(gzipcs); + cs->private_data = NULL; +} + +static void +WriteDataToArchiveGzip(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen) +{ + GzipCompressorState *gzipcs = (GzipCompressorState *) cs->private_data; + z_streamp zp; + + if (!gzipcs->zp) + { + zp = gzipcs->zp = (z_streamp) pg_malloc(sizeof(z_stream)); + zp->zalloc = Z_NULL; + zp->zfree = Z_NULL; + zp->opaque = Z_NULL; + + /* + * outsize is the buffer size we tell zlib it can output to. We + * actually allocate one extra byte because some routines want to + * append a trailing zero byte to the zlib output. + */ + gzipcs->outbuf = pg_malloc(ZLIB_OUT_SIZE + 1); + gzipcs->outsize = ZLIB_OUT_SIZE; + + /* + * A level of zero simply copies the input one block at the time. This + * is probably not what the user wanted when calling this interface. + */ + if (cs->compression_spec.level == 0) + pg_fatal("requested to compress the archive yet no level was specified"); + + if (deflateInit(zp, cs->compression_spec.level) != Z_OK) + pg_fatal("could not initialize compression library: %s", zp->msg); + + /* Just be paranoid - maybe End is called after Start, with no Write */ + zp->next_out = gzipcs->outbuf; + zp->avail_out = gzipcs->outsize; + } + + gzipcs->zp->next_in = (void *) unconstify(void *, data); + gzipcs->zp->avail_in = dLen; + DeflateCompressorGzip(AH, cs, false); +} + +static void +ReadDataFromArchiveGzip(ArchiveHandle *AH, CompressorState *cs) +{ + z_streamp zp; + char *out; + int res = Z_OK; + size_t cnt; + char *buf; + size_t buflen; + + zp = (z_streamp) pg_malloc(sizeof(z_stream)); + zp->zalloc = Z_NULL; + zp->zfree = Z_NULL; + zp->opaque = Z_NULL; + + buf = pg_malloc(ZLIB_IN_SIZE); + buflen = ZLIB_IN_SIZE; + + out = pg_malloc(ZLIB_OUT_SIZE + 1); + + if (inflateInit(zp) != Z_OK) + pg_fatal("could not initialize compression library: %s", + zp->msg); + + /* no minimal chunk size for zlib */ + while ((cnt = cs->readF(AH, &buf, &buflen))) + { + zp->next_in = (void *) buf; + zp->avail_in = cnt; + + while (zp->avail_in > 0) + { + zp->next_out = (void *) out; + zp->avail_out = ZLIB_OUT_SIZE; + + res = inflate(zp, 0); + if (res != Z_OK && res != Z_STREAM_END) + pg_fatal("could not uncompress data: %s", zp->msg); + + out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; + ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); + } + } + + zp->next_in = NULL; + zp->avail_in = 0; + while (res != Z_STREAM_END) + { + zp->next_out = (void *) out; + zp->avail_out = ZLIB_OUT_SIZE; + res = inflate(zp, 0); + if (res != Z_OK && res != Z_STREAM_END) + pg_fatal("could not uncompress data: %s", zp->msg); + + out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; + ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); + } + + if (inflateEnd(zp) != Z_OK) + pg_fatal("could not close compression library: %s", zp->msg); + + free(buf); + free(out); + free(zp); +} + +/* Public routines that support gzip compressed data I/O */ +void +InitCompressorGzip(CompressorState *cs, const pg_compress_specification *compression_spec) +{ + GzipCompressorState *gzipcs; + + cs->readData = ReadDataFromArchiveGzip; + cs->writeData = WriteDataToArchiveGzip; + cs->end = EndCompressorGzip; + + cs->compression_spec = *compression_spec; + + gzipcs = (GzipCompressorState *) pg_malloc0(sizeof(GzipCompressorState)); + + cs->private_data = gzipcs; +} + + +/*---------------------- + * Compress File API + *---------------------- + */ + +static size_t +Gzip_read(void *ptr, size_t size, CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + size_t ret; + + ret = gzread(gzfp, ptr, size); + if (ret != size && !gzeof(gzfp)) + { + int errnum; + const char *errmsg = gzerror(gzfp, &errnum); + + pg_fatal("could not read from input file: %s", + errnum == Z_ERRNO ? strerror(errno) : errmsg); + } + + return ret; +} + +static size_t +Gzip_write(const void *ptr, size_t size, CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + + return gzwrite(gzfp, ptr, size); +} + +static int +Gzip_getc(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + int ret; + + errno = 0; + ret = gzgetc(gzfp); + if (ret == EOF) + { + if (!gzeof(gzfp)) + pg_fatal("could not read from input file: %s", strerror(errno)); + else + pg_fatal("could not read from input file: end of file"); + } + + return ret; +} + +static char * +Gzip_gets(char *ptr, int size, CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + + return gzgets(gzfp, ptr, size); +} + +static int +Gzip_close(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + int save_errno; + int ret; + + CFH->private_data = NULL; + + ret = gzclose(gzfp); + + save_errno = errno; + errno = save_errno; + + return ret; +} + +static int +Gzip_eof(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + + return gzeof(gzfp); +} + +static const char * +Gzip_get_error(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + const char *errmsg; + int errnum; + + errmsg = gzerror(gzfp, &errnum); + if (errnum == Z_ERRNO) + errmsg = strerror(errno); + + return errmsg; +} + +static int +Gzip_open(const char *path, int fd, const char *mode, CompressFileHandle *CFH) +{ + gzFile gzfp; + char mode_compression[32]; + + if (CFH->compression_spec.level != Z_DEFAULT_COMPRESSION) + { + /* + * user has specified a compression level, so tell zlib to use it + */ + snprintf(mode_compression, sizeof(mode_compression), "%s%d", + mode, CFH->compression_spec.level); + } + else + strcpy(mode_compression, mode); + + if (fd >= 0) + gzfp = gzdopen(dup(fd), mode_compression); + else + gzfp = gzopen(path, mode_compression); + + if (gzfp == NULL) + return 1; + + CFH->private_data = gzfp; + + return 0; +} + +static int +Gzip_open_write(const char *path, const char *mode, CompressFileHandle *CFH) +{ + char *fname; + int ret; + int save_errno; + + fname = psprintf("%s.gz", path); + ret = CFH->open_func(fname, -1, mode, CFH); + + save_errno = errno; + pg_free(fname); + errno = save_errno; + + return ret; +} + +void +InitCompressGzip(CompressFileHandle *CFH, const pg_compress_specification *compression_spec) +{ + CFH->open_func = Gzip_open; + CFH->open_write_func = Gzip_open_write; + CFH->read_func = Gzip_read; + CFH->write_func = Gzip_write; + CFH->gets_func = Gzip_gets; + CFH->getc_func = Gzip_getc; + CFH->close_func = Gzip_close; + CFH->eof_func = Gzip_eof; + CFH->get_error_func = Gzip_get_error; + + CFH->compression_spec = *compression_spec; + + CFH->private_data = NULL; +} +#else /* HAVE_LIBZ */ +void +InitCompressorGzip(CompressorState *cs, const pg_compress_specification *compression_spec) +{ + pg_fatal("this build does not support compression with %s", "gzip"); +} + +void +InitCompressGzip(CompressFileHandle *CFH, const pg_compress_specification *compression_spec) +{ + pg_fatal("this build does not support compression with %s", "gzip"); +} +#endif /* HAVE_LIBZ */ diff --git a/src/bin/pg_dump/compress_gzip.h b/src/bin/pg_dump/compress_gzip.h new file mode 100644 index 00000000000..a1fc3595e51 --- /dev/null +++ b/src/bin/pg_dump/compress_gzip.h @@ -0,0 +1,22 @@ +/*------------------------------------------------------------------------- + * + * compress_gzip.h + * Interface to compress_io.c routines + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_gzip.h + * + *------------------------------------------------------------------------- + */ +#ifndef _COMPRESS_GZIP_H_ +#define _COMPRESS_GZIP_H_ + +#include "compress_io.h" + +extern void InitCompressorGzip(CompressorState *cs, const pg_compress_specification *compression_spec); +extern void InitCompressGzip(CompressFileHandle *CFH, const pg_compress_specification *compression_spec); + +#endif /* _COMPRESS_GZIP_H_ */ diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index 8d0bec08d7c..c4ac4042794 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -9,42 +9,44 @@ * * This file includes two APIs for dealing with compressed data. The first * provides more flexibility, using callbacks to read/write data from the - * underlying stream. The second API is a wrapper around fopen/gzopen and + * underlying stream. The second API is a wrapper around fopen and * friends, providing an interface similar to those, but abstracts away - * the possible compression. Both APIs use libz for the compression, but - * the second API uses gzip headers, so the resulting files can be easily - * manipulated with the gzip utility. + * the possible compression. The second API is aimed for the resulting + * files can be easily manipulated with an external compression utility + * program. * * Compressor API * -------------- * * The interface for writing to an archive consists of three functions: - * AllocateCompressor, WriteDataToArchive and EndCompressor. First you call - * AllocateCompressor, then write all the data by calling WriteDataToArchive - * as many times as needed, and finally EndCompressor. WriteDataToArchive - * and EndCompressor will call the WriteFunc that was provided to - * AllocateCompressor for each chunk of compressed data. + * AllocateCompressor, writeData, and EndCompressor. First you call + * AllocateCompressor, then write all the data by calling writeData as many + * times as needed, and finally EndCompressor. writeData will call the + * WriteFunc that was provided to AllocateCompressor for each chunk of + * compressed data. * - * The interface for reading an archive consists of just one function: - * ReadDataFromArchive. ReadDataFromArchive reads the whole compressed input - * stream, by repeatedly calling the given ReadFunc. ReadFunc returns the - * compressed data chunk at a time, and ReadDataFromArchive decompresses it - * and passes the decompressed data to ahwrite(), until ReadFunc returns 0 - * to signal EOF. - * - * The interface is the same for compressed and uncompressed streams. + * The interface for reading an archive consists of the same three functions: + * AllocateCompressor, readData, and EndCompressor. First you call + * AllocateCompressor, then read all the data by calling readData to read the + * whole compressed stream which repeatedly calls the given ReadFunc. ReadFunc + * returns the compressed data chunk at a time, and readData decompresses it + * and passes the decompressed data to ahwrite(), until ReadFunc returns 0 to + * signal EOF. The interface is the same for compressed and uncompressed + * streams. * * Compressed stream API * ---------------------- * * The compressed stream API is a wrapper around the C standard fopen() and - * libz's gzopen() APIs. It allows you to use the same functions for - * compressed and uncompressed streams. cfopen_read() first tries to open - * the file with given name, and if it fails, it tries to open the same - * file with the .gz suffix. cfopen_write() opens a file for writing, an - * extra argument specifies if the file should be compressed, and adds the - * .gz suffix to the filename if so. This allows you to easily handle both - * compressed and uncompressed files. + * libz's gzopen() APIs and custom LZ4 calls which provide similar + * functionality. It allows you to use the same functions for compressed and + * uncompressed streams. cfopen_read() first tries to open the file with given + * name, and if it fails, it tries to open the same file with the .gz suffix, + * failing that it tries to open the same file with the .lz4 suffix. + * cfopen_write() opens a file for writing, an extra argument specifies the + * method to use should the file be compressed, and adds the appropriate + * suffix, .gz or .lz4, to the filename if so. This allows you to easily handle + * both compressed and uncompressed files. * * IDENTIFICATION * src/bin/pg_dump/compress_io.c @@ -53,7 +55,11 @@ */ #include "postgres_fe.h" +#include <sys/stat.h> +#include <unistd.h> + #include "compress_io.h" +#include "compress_gzip.h" #include "pg_backup_utils.h" #ifdef HAVE_LIBZ @@ -65,84 +71,70 @@ *---------------------- */ -/* typedef appears in compress_io.h */ -struct CompressorState +static void +ReadDataFromArchiveNone(ArchiveHandle *AH, CompressorState *cs) { - pg_compress_specification compression_spec; - WriteFunc writeF; + size_t cnt; + char *buf; + size_t buflen; -#ifdef HAVE_LIBZ - z_streamp zp; - char *zlibOut; - size_t zlibOutSize; -#endif -}; + buf = pg_malloc(ZLIB_OUT_SIZE); + buflen = ZLIB_OUT_SIZE; + + while ((cnt = cs->readF(AH, &buf, &buflen))) + { + ahwrite(buf, 1, cnt, AH); + } + + free(buf); +} -/* Routines that support zlib compressed data I/O */ -#ifdef HAVE_LIBZ -static void InitCompressorZlib(CompressorState *cs, int level); -static void DeflateCompressorZlib(ArchiveHandle *AH, CompressorState *cs, - bool flush); -static void ReadDataFromArchiveZlib(ArchiveHandle *AH, ReadFunc readF); -static void WriteDataToArchiveZlib(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen); -static void EndCompressorZlib(ArchiveHandle *AH, CompressorState *cs); -#endif -/* Routines that support uncompressed data I/O */ -static void ReadDataFromArchiveNone(ArchiveHandle *AH, ReadFunc readF); -static void WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen); +static void +WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen) +{ + cs->writeF(AH, data, dLen); +} + +static void +EndCompressorNone(ArchiveHandle *AH, CompressorState *cs) +{ + /* no op */ +} + +static void +InitCompressorNone(CompressorState *cs, + const pg_compress_specification *compression_spec) +{ + cs->readData = ReadDataFromArchiveNone; + cs->writeData = WriteDataToArchiveNone; + cs->end = EndCompressorNone; + + cs->compression_spec = *compression_spec; +} /* Public interface routines */ /* Allocate a new compressor */ CompressorState * AllocateCompressor(const pg_compress_specification *compression_spec, - WriteFunc writeF) + ReadFunc readF, WriteFunc writeF) { CompressorState *cs; -#ifndef HAVE_LIBZ - if (compression_spec->algorithm == PG_COMPRESSION_GZIP) - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - cs = (CompressorState *) pg_malloc0(sizeof(CompressorState)); + cs->readF = readF; cs->writeF = writeF; cs->compression_spec = *compression_spec; - /* - * Perform compression algorithm specific initialization. - */ -#ifdef HAVE_LIBZ - if (cs->compression_spec.algorithm == PG_COMPRESSION_GZIP) - InitCompressorZlib(cs, cs->compression_spec.level); -#endif - - return cs; -} - -/* - * Read all compressed data from the input stream (via readF) and print it - * out with ahwrite(). - */ -void -ReadDataFromArchive(ArchiveHandle *AH, - const pg_compress_specification *compression_spec, - ReadFunc readF) -{ switch (compression_spec->algorithm) { case PG_COMPRESSION_NONE: - ReadDataFromArchiveNone(AH, readF); + InitCompressorNone(cs, compression_spec); break; case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ReadDataFromArchiveZlib(AH, readF); -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif + InitCompressorGzip(cs, compression_spec); break; case PG_COMPRESSION_LZ4: pg_fatal("compression with %s is not yet supported", "LZ4"); @@ -151,34 +143,8 @@ ReadDataFromArchive(ArchiveHandle *AH, pg_fatal("compression with %s is not yet supported", "ZSTD"); break; } -} -/* - * Compress and write data to the output stream (via writeF). - */ -void -WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, - const void *data, size_t dLen) -{ - switch (cs->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - WriteDataToArchiveNone(AH, cs, data, dLen); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - WriteDataToArchiveZlib(AH, cs, data, dLen); -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } + return cs; } /* @@ -187,401 +153,177 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, void EndCompressor(ArchiveHandle *AH, CompressorState *cs) { - switch (cs->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - EndCompressorZlib(AH, cs); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - free(cs); + cs->end(AH, cs); + pg_free(cs); } -/* Private routines, specific to each compression method. */ - -#ifdef HAVE_LIBZ -/* - * Functions for zlib compressed output. +/*---------------------- + * Compressed stream API + *---------------------- */ -static void -InitCompressorZlib(CompressorState *cs, int level) +static int +hasSuffix(const char *filename, const char *suffix) { - z_streamp zp; - - zp = cs->zp = (z_streamp) pg_malloc(sizeof(z_stream)); - zp->zalloc = Z_NULL; - zp->zfree = Z_NULL; - zp->opaque = Z_NULL; - - /* - * zlibOutSize is the buffer size we tell zlib it can output to. We - * actually allocate one extra byte because some routines want to append a - * trailing zero byte to the zlib output. - */ - cs->zlibOut = (char *) pg_malloc(ZLIB_OUT_SIZE + 1); - cs->zlibOutSize = ZLIB_OUT_SIZE; - - if (deflateInit(zp, level) != Z_OK) - pg_fatal("could not initialize compression library: %s", - zp->msg); - - /* Just be paranoid - maybe End is called after Start, with no Write */ - zp->next_out = (void *) cs->zlibOut; - zp->avail_out = cs->zlibOutSize; + int filenamelen = strlen(filename); + int suffixlen = strlen(suffix); + + if (filenamelen < suffixlen) + return 0; + + return memcmp(&filename[filenamelen - suffixlen], + suffix, + suffixlen) == 0; } +/* free() without changing errno; useful in several places below */ static void -EndCompressorZlib(ArchiveHandle *AH, CompressorState *cs) +free_keep_errno(void *p) { - z_streamp zp = cs->zp; - - zp->next_in = NULL; - zp->avail_in = 0; - - /* Flush any remaining data from zlib buffer */ - DeflateCompressorZlib(AH, cs, true); - - if (deflateEnd(zp) != Z_OK) - pg_fatal("could not close compression stream: %s", zp->msg); + int save_errno = errno; - free(cs->zlibOut); - free(cs->zp); + free(p); + errno = save_errno; } -static void -DeflateCompressorZlib(ArchiveHandle *AH, CompressorState *cs, bool flush) +/* + * Compression None implementation + */ +static size_t +read_none(void *ptr, size_t size, CompressFileHandle *CFH) { - z_streamp zp = cs->zp; - char *out = cs->zlibOut; - int res = Z_OK; + FILE *fp = (FILE *) CFH->private_data; + size_t ret; - while (cs->zp->avail_in != 0 || flush) - { - res = deflate(zp, flush ? Z_FINISH : Z_NO_FLUSH); - if (res == Z_STREAM_ERROR) - pg_fatal("could not compress data: %s", zp->msg); - if ((flush && (zp->avail_out < cs->zlibOutSize)) - || (zp->avail_out == 0) - || (zp->avail_in != 0) - ) - { - /* - * Extra paranoia: avoid zero-length chunks, since a zero length - * chunk is the EOF marker in the custom format. This should never - * happen but... - */ - if (zp->avail_out < cs->zlibOutSize) - { - /* - * Any write function should do its own error checking but to - * make sure we do a check here as well... - */ - size_t len = cs->zlibOutSize - zp->avail_out; - - cs->writeF(AH, out, len); - } - zp->next_out = (void *) out; - zp->avail_out = cs->zlibOutSize; - } + if (size == 0) + return 0; - if (res == Z_STREAM_END) - break; - } + ret = fread(ptr, 1, size, fp); + if (ret != size && !feof(fp)) + pg_fatal("could not read from input file: %s", + strerror(errno)); + + return ret; } -static void -WriteDataToArchiveZlib(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen) +static size_t +write_none(const void *ptr, size_t size, CompressFileHandle *CFH) { - cs->zp->next_in = (void *) unconstify(char *, data); - cs->zp->avail_in = dLen; - DeflateCompressorZlib(AH, cs, false); + return fwrite(ptr, 1, size, (FILE *) CFH->private_data); } -static void -ReadDataFromArchiveZlib(ArchiveHandle *AH, ReadFunc readF) +static const char * +get_error_none(CompressFileHandle *CFH) { - z_streamp zp; - char *out; - int res = Z_OK; - size_t cnt; - char *buf; - size_t buflen; - - zp = (z_streamp) pg_malloc(sizeof(z_stream)); - zp->zalloc = Z_NULL; - zp->zfree = Z_NULL; - zp->opaque = Z_NULL; - - buf = pg_malloc(ZLIB_IN_SIZE); - buflen = ZLIB_IN_SIZE; - - out = pg_malloc(ZLIB_OUT_SIZE + 1); - - if (inflateInit(zp) != Z_OK) - pg_fatal("could not initialize compression library: %s", - zp->msg); - - /* no minimal chunk size for zlib */ - while ((cnt = readF(AH, &buf, &buflen))) - { - zp->next_in = (void *) buf; - zp->avail_in = cnt; - - while (zp->avail_in > 0) - { - zp->next_out = (void *) out; - zp->avail_out = ZLIB_OUT_SIZE; + return strerror(errno); +} - res = inflate(zp, 0); - if (res != Z_OK && res != Z_STREAM_END) - pg_fatal("could not uncompress data: %s", zp->msg); +static char * +gets_none(char *ptr, int size, CompressFileHandle *CFH) +{ + return fgets(ptr, size, (FILE *) CFH->private_data); +} - out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; - ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); - } - } +static int +getc_none(CompressFileHandle *CFH) +{ + FILE *fp = (FILE *) CFH->private_data; + int ret; - zp->next_in = NULL; - zp->avail_in = 0; - while (res != Z_STREAM_END) + ret = fgetc(fp); + if (ret == EOF) { - zp->next_out = (void *) out; - zp->avail_out = ZLIB_OUT_SIZE; - res = inflate(zp, 0); - if (res != Z_OK && res != Z_STREAM_END) - pg_fatal("could not uncompress data: %s", zp->msg); - - out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; - ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); + if (!feof(fp)) + pg_fatal("could not read from input file: %s", strerror(errno)); + else + pg_fatal("could not read from input file: end of file"); } - if (inflateEnd(zp) != Z_OK) - pg_fatal("could not close compression library: %s", zp->msg); - - free(buf); - free(out); - free(zp); + return ret; } -#endif /* HAVE_LIBZ */ - - -/* - * Functions for uncompressed output. - */ -static void -ReadDataFromArchiveNone(ArchiveHandle *AH, ReadFunc readF) +static int +close_none(CompressFileHandle *CFH) { - size_t cnt; - char *buf; - size_t buflen; + FILE *fp = (FILE *) CFH->private_data; + int ret = 0; - buf = pg_malloc(ZLIB_OUT_SIZE); - buflen = ZLIB_OUT_SIZE; + CFH->private_data = NULL; - while ((cnt = readF(AH, &buf, &buflen))) - { - ahwrite(buf, 1, cnt, AH); - } + if (fp) + ret = fclose(fp); - free(buf); + return ret; } -static void -WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen) + +static int +eof_none(CompressFileHandle *CFH) { - cs->writeF(AH, data, dLen); + return feof((FILE *) CFH->private_data); } - -/*---------------------- - * Compressed stream API - *---------------------- - */ - -/* - * cfp represents an open stream, wrapping the underlying FILE or gzFile - * pointer. This is opaque to the callers. - */ -struct cfp +static int +open_none(const char *path, int fd, const char *mode, CompressFileHandle *CFH) { - pg_compress_specification compression_spec; - void *fp; -}; + Assert(CFH->private_data == NULL); -#ifdef HAVE_LIBZ -static int hasSuffix(const char *filename, const char *suffix); -#endif + if (fd >= 0) + CFH->private_data = fdopen(dup(fd), mode); + else + CFH->private_data = fopen(path, mode); -/* free() without changing errno; useful in several places below */ -static void -free_keep_errno(void *p) -{ - int save_errno = errno; + if (CFH->private_data == NULL) + return 1; - free(p); - errno = save_errno; + return 0; } -/* - * Open a file for reading. 'path' is the file to open, and 'mode' should - * be either "r" or "rb". - * - * If the file at 'path' does not exist, we append the ".gz" suffix (if 'path' - * doesn't already have it) and try again. So if you pass "foo" as 'path', - * this will open either "foo" or "foo.gz". - * - * On failure, return NULL with an error code in errno. - */ -cfp * -cfopen_read(const char *path, const char *mode) +static int +open_write_none(const char *path, const char *mode, CompressFileHandle *CFH) { - cfp *fp; - - pg_compress_specification compression_spec = {0}; + Assert(CFH->private_data == NULL); + CFH->private_data = fopen(path, mode); + if (CFH->private_data == NULL) + return 1; -#ifdef HAVE_LIBZ - if (hasSuffix(path, ".gz")) - { - compression_spec.algorithm = PG_COMPRESSION_GZIP; - fp = cfopen(path, mode, &compression_spec); - } - else -#endif - { - compression_spec.algorithm = PG_COMPRESSION_NONE; - fp = cfopen(path, mode, &compression_spec); -#ifdef HAVE_LIBZ - if (fp == NULL) - { - char *fname; - - fname = psprintf("%s.gz", path); - compression_spec.algorithm = PG_COMPRESSION_GZIP; - fp = cfopen(fname, mode, &compression_spec); - free_keep_errno(fname); - } -#endif - } - return fp; + return 0; } -/* - * Open a file for writing. 'path' indicates the path name, and 'mode' must - * be a filemode as accepted by fopen() and gzopen() that indicates writing - * ("w", "wb", "a", or "ab"). - * - * If 'compression_spec.algorithm' is GZIP, a gzip compressed stream is opened, - * and 'compression_spec.level' used. The ".gz" suffix is automatically added to - * 'path' in that case. - * - * On failure, return NULL with an error code in errno. - */ -cfp * -cfopen_write(const char *path, const char *mode, - const pg_compress_specification *compression_spec) +static void +InitCompressNone(CompressFileHandle *CFH, + const pg_compress_specification *compression_spec) { - cfp *fp; - - if (compression_spec->algorithm == PG_COMPRESSION_NONE) - fp = cfopen(path, mode, compression_spec); - else - { -#ifdef HAVE_LIBZ - char *fname; - - fname = psprintf("%s.gz", path); - fp = cfopen(fname, mode, compression_spec); - free_keep_errno(fname); -#else - pg_fatal("this build does not support compression with %s", "gzip"); - fp = NULL; /* keep compiler quiet */ -#endif - } - return fp; + CFH->open_func = open_none; + CFH->open_write_func = open_write_none; + CFH->read_func = read_none; + CFH->write_func = write_none; + CFH->gets_func = gets_none; + CFH->getc_func = getc_none; + CFH->close_func = close_none; + CFH->eof_func = eof_none; + CFH->get_error_func = get_error_none; + + CFH->private_data = NULL; } /* - * This is the workhorse for cfopen() or cfdopen(). It opens file 'path' or - * associates a stream 'fd', if 'fd' is a valid descriptor, in 'mode'. The - * descriptor is not dup'ed and it is the caller's responsibility to do so. - * The caller must verify that the 'compress_algorithm' is supported by the - * current build. - * - * On failure, return NULL with an error code in errno. + * Public interface */ -static cfp * -cfopen_internal(const char *path, int fd, const char *mode, - const pg_compress_specification *compression_spec) +CompressFileHandle * +InitCompressFileHandle(const pg_compress_specification *compression_spec) { - cfp *fp = pg_malloc(sizeof(cfp)); + CompressFileHandle *CFH; - fp->compression_spec = *compression_spec; + CFH = pg_malloc0(sizeof(CompressFileHandle)); switch (compression_spec->algorithm) { case PG_COMPRESSION_NONE: - if (fd >= 0) - fp->fp = fdopen(fd, mode); - else - fp->fp = fopen(path, mode); - if (fp->fp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } - + InitCompressNone(CFH, compression_spec); break; case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - if (compression_spec->level != Z_DEFAULT_COMPRESSION) - { - /* - * user has specified a compression level, so tell zlib to use - * it - */ - char mode_compression[32]; - - snprintf(mode_compression, sizeof(mode_compression), "%s%d", - mode, compression_spec->level); - if (fd >= 0) - fp->fp = gzdopen(fd, mode_compression); - else - fp->fp = gzopen(path, mode_compression); - } - else - { - /* don't specify a level, just use the zlib default */ - if (fd >= 0) - fp->fp = gzdopen(fd, mode); - else - fp->fp = gzopen(path, mode); - } - - if (fp->fp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif + InitCompressGzip(CFH, compression_spec); break; case PG_COMPRESSION_LZ4: pg_fatal("compression with %s is not yet supported", "LZ4"); @@ -591,266 +333,88 @@ cfopen_internal(const char *path, int fd, const char *mode, break; } - return fp; + return CFH; } -cfp * -cfopen(const char *path, const char *mode, - const pg_compress_specification *compression_spec) +/* + * Open a file for reading. 'path' is the file to open, and 'mode' should + * be either "r" or "rb". + * + * If the file at 'path' does not exist, we append the "{.gz,.lz4}" suffix (i + * 'path' doesn't already have it) and try again. So if you pass "foo" as + * 'path', this will open either "foo" or "foo.gz" or "foo.lz4", trying in that + * order. + * + * On failure, return NULL with an error code in errno. + */ +CompressFileHandle * +InitDiscoverCompressFileHandle(const char *path, const char *mode) { - return cfopen_internal(path, -1, mode, compression_spec); -} + CompressFileHandle *CFH = NULL; + struct stat st; + char *fname; + pg_compress_specification compression_spec = {0}; -cfp * -cfdopen(int fd, const char *mode, - const pg_compress_specification *compression_spec) -{ - return cfopen_internal(NULL, fd, mode, compression_spec); -} + // compression_spec.algorithm = PG_COMPRESSION_NONE; -int -cfread(void *ptr, int size, cfp *fp) -{ - int ret = 0; + Assert(strcmp(mode, "r") == 0 || strcmp(mode, "rb") == 0); - if (size == 0) - return 0; + fname = strdup(path); - switch (fp->compression_spec.algorithm) + if (hasSuffix(fname, ".gz")) + compression_spec.algorithm = PG_COMPRESSION_GZIP; + else { - case PG_COMPRESSION_NONE: - ret = fread(ptr, 1, size, (FILE *) fp->fp); - if (ret != size && !feof((FILE *) fp->fp)) - READ_ERROR_EXIT((FILE *) fp->fp); + bool exists; - break; - case PG_COMPRESSION_GZIP: + exists = (stat(path, &st) == 0); + /* avoid unused warning if it is not build with compression */ + if (exists) + compression_spec.algorithm = PG_COMPRESSION_NONE; #ifdef HAVE_LIBZ - ret = gzread((gzFile) fp->fp, ptr, size); - if (ret != size && !gzeof((gzFile) fp->fp)) - { - int errnum; - const char *errmsg = gzerror((gzFile) fp->fp, &errnum); - - pg_fatal("could not read from input file: %s", - errnum == Z_ERRNO ? strerror(errno) : errmsg); - } -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - return ret; -} - -int -cfwrite(const void *ptr, int size, cfp *fp) -{ - int ret = 0; - - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = fwrite(ptr, 1, size, (FILE *) fp->fp); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzwrite((gzFile) fp->fp, ptr, size); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - return ret; -} - -int -cfgetc(cfp *fp) -{ - int ret = 0; - - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = fgetc((FILE *) fp->fp); - if (ret == EOF) - READ_ERROR_EXIT((FILE *) fp->fp); + if (!exists) + { + free_keep_errno(fname); + fname = psprintf("%s.gz", path); + exists = (stat(fname, &st) == 0); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzgetc((gzFile) fp->fp); - if (ret == EOF) - { - if (!gzeof((gzFile) fp->fp)) - pg_fatal("could not read from input file: %s", strerror(errno)); - else - pg_fatal("could not read from input file: end of file"); - } -#else - pg_fatal("this build does not support compression with %s", - "gzip"); + if (exists) + compression_spec.algorithm = PG_COMPRESSION_GZIP; + } #endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - return ret; -} - -char * -cfgets(cfp *fp, char *buf, int len) -{ - char *ret = NULL; - - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = fgets(buf, len, (FILE *) fp->fp); +#ifdef USE_LZ4 + if (!exists) + { + free_keep_errno(fname); + fname = psprintf("%s.lz4", path); + exists = (stat(fname, &st) == 0); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzgets((gzFile) fp->fp, buf, len); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); + if (exists) + compression_spec.algorithm = PG_COMPRESSION_LZ4; + } #endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - return ret; -} - -int -cfclose(cfp *fp) -{ - int ret = 0; - - if (fp == NULL) - { - errno = EBADF; - return EOF; } - switch (fp->compression_spec.algorithm) + CFH = InitCompressFileHandle(&compression_spec); + if (CFH->open_func(fname, -1, mode, CFH)) { - case PG_COMPRESSION_NONE: - ret = fclose((FILE *) fp->fp); - fp->fp = NULL; - - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzclose((gzFile) fp->fp); - fp->fp = NULL; -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; + free_keep_errno(CFH); + CFH = NULL; } + free_keep_errno(fname); - free_keep_errno(fp); - - return ret; + return CFH; } int -cfeof(cfp *fp) +DestroyCompressFileHandle(CompressFileHandle *CFH) { int ret = 0; - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = feof((FILE *) fp->fp); + if (CFH->private_data) + ret = CFH->close_func(CFH); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzeof((gzFile) fp->fp); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } + free_keep_errno(CFH); return ret; } - -const char * -get_cfp_error(cfp *fp) -{ - if (fp->compression_spec.algorithm == PG_COMPRESSION_GZIP) - { -#ifdef HAVE_LIBZ - int errnum; - const char *errmsg = gzerror((gzFile) fp->fp, &errnum); - - if (errnum != Z_ERRNO) - return errmsg; -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - } - - return strerror(errno); -} - -#ifdef HAVE_LIBZ -static int -hasSuffix(const char *filename, const char *suffix) -{ - int filenamelen = strlen(filename); - int suffixlen = strlen(suffix); - - if (filenamelen < suffixlen) - return 0; - - return memcmp(&filename[filenamelen - suffixlen], - suffix, - suffixlen) == 0; -} - -#endif diff --git a/src/bin/pg_dump/compress_io.h b/src/bin/pg_dump/compress_io.h index 768096c820d..afe6b22efaf 100644 --- a/src/bin/pg_dump/compress_io.h +++ b/src/bin/pg_dump/compress_io.h @@ -37,34 +37,63 @@ typedef void (*WriteFunc) (ArchiveHandle *AH, const char *buf, size_t len); */ typedef size_t (*ReadFunc) (ArchiveHandle *AH, char **buf, size_t *buflen); -/* struct definition appears in compress_io.c */ typedef struct CompressorState CompressorState; +struct CompressorState +{ + /* + * Read all compressed data from the input stream (via readF) and print it + * out with ahwrite(). + */ + void (*readData) (ArchiveHandle *AH, CompressorState *cs); + + /* + * Compress and write data to the output stream (via writeF). + */ + void (*writeData) (ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen); + void (*end) (ArchiveHandle *AH, CompressorState *cs); + + ReadFunc readF; + WriteFunc writeF; + + pg_compress_specification compression_spec; + void *private_data; +}; extern CompressorState *AllocateCompressor(const pg_compress_specification *compression_spec, + ReadFunc readF, WriteFunc writeF); -extern void ReadDataFromArchive(ArchiveHandle *AH, - const pg_compress_specification *compression_spec, - ReadFunc readF); -extern void WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, - const void *data, size_t dLen); extern void EndCompressor(ArchiveHandle *AH, CompressorState *cs); +/* + * Compress File Handle + */ +typedef struct CompressFileHandle CompressFileHandle; + +struct CompressFileHandle +{ + int (*open_func) (const char *path, int fd, const char *mode, + CompressFileHandle *CFH); + int (*open_write_func) (const char *path, const char *mode, + CompressFileHandle *cxt); + size_t (*read_func) (void *ptr, size_t size, CompressFileHandle *CFH); + size_t (*write_func) (const void *ptr, size_t size, + struct CompressFileHandle *CFH); + char *(*gets_func) (char *s, int size, CompressFileHandle *CFH); + int (*getc_func) (CompressFileHandle *CFH); + int (*eof_func) (CompressFileHandle *CFH); + int (*close_func) (CompressFileHandle *CFH); + const char *(*get_error_func) (CompressFileHandle *CFH); + + pg_compress_specification compression_spec; + void *private_data; +}; -typedef struct cfp cfp; +extern CompressFileHandle *InitCompressFileHandle( + const pg_compress_specification *compression_spec); -extern cfp *cfopen(const char *path, const char *mode, - const pg_compress_specification *compression_spec); -extern cfp *cfdopen(int fd, const char *mode, - const pg_compress_specification *compression_spec); -extern cfp *cfopen_read(const char *path, const char *mode); -extern cfp *cfopen_write(const char *path, const char *mode, - const pg_compress_specification *compression_spec); -extern int cfread(void *ptr, int size, cfp *fp); -extern int cfwrite(const void *ptr, int size, cfp *fp); -extern int cfgetc(cfp *fp); -extern char *cfgets(cfp *fp, char *buf, int len); -extern int cfclose(cfp *fp); -extern int cfeof(cfp *fp); -extern const char *get_cfp_error(cfp *fp); +extern CompressFileHandle *InitDiscoverCompressFileHandle(const char *path, + const char *mode); +extern int DestroyCompressFileHandle(CompressFileHandle *CFH); #endif diff --git a/src/bin/pg_dump/meson.build b/src/bin/pg_dump/meson.build index ca62f9a3740..aa2c91829c0 100644 --- a/src/bin/pg_dump/meson.build +++ b/src/bin/pg_dump/meson.build @@ -2,6 +2,7 @@ pg_dump_common_sources = files( 'compress_io.c', + 'compress_gzip.c', 'dumputils.c', 'parallel.c', 'pg_backup_archiver.c', diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index d4f1e09fce6..9e9f1d626b5 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -95,8 +95,8 @@ static void dump_lo_buf(ArchiveHandle *AH); static void dumpTimestamp(ArchiveHandle *AH, const char *msg, time_t tim); static void SetOutput(ArchiveHandle *AH, const char *filename, const pg_compress_specification *compression_spec); -static cfp *SaveOutput(ArchiveHandle *AH); -static void RestoreOutput(ArchiveHandle *AH, cfp *savedOutput); +static CompressFileHandle *SaveOutput(ArchiveHandle *AH); +static void RestoreOutput(ArchiveHandle *AH, CompressFileHandle *savedOutput); static int restore_toc_entry(ArchiveHandle *AH, TocEntry *te, bool is_parallel); static void restore_toc_entries_prefork(ArchiveHandle *AH, @@ -272,7 +272,7 @@ CloseArchive(Archive *AHX) /* Close the output */ errno = 0; - res = cfclose(AH->OF); + res = DestroyCompressFileHandle(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); @@ -355,7 +355,7 @@ RestoreArchive(Archive *AHX) bool parallel_mode; bool supports_compression; TocEntry *te; - cfp *sav; + CompressFileHandle *sav; AH->stage = STAGE_INITIALIZING; @@ -1126,7 +1126,7 @@ PrintTOCSummary(Archive *AHX) TocEntry *te; pg_compress_specification out_compression_spec = {0}; teSection curSection; - cfp *sav; + CompressFileHandle *sav; const char *fmtName; char stamp_str[64]; @@ -1142,9 +1142,10 @@ PrintTOCSummary(Archive *AHX) strcpy(stamp_str, "[unknown]"); ahprintf(AH, ";\n; Archive created at %s\n", stamp_str); - ahprintf(AH, "; dbname: %s\n; TOC Entries: %d\n; Compression: %d\n", + ahprintf(AH, "; dbname: %s\n; TOC Entries: %d\n; Compression: %s\n", sanitize_line(AH->archdbname, false), - AH->tocCount, AH->compression_spec.level); + AH->tocCount, + get_compress_algorithm_name(AH->compression_spec.algorithm)); switch (AH->format) { @@ -1501,6 +1502,7 @@ static void SetOutput(ArchiveHandle *AH, const char *filename, const pg_compress_specification *compression_spec) { + CompressFileHandle *CFH; const char *mode; int fn = -1; @@ -1523,33 +1525,32 @@ SetOutput(ArchiveHandle *AH, const char *filename, else mode = PG_BINARY_W; - if (fn >= 0) - AH->OF = cfdopen(dup(fn), mode, compression_spec); - else - AH->OF = cfopen(filename, mode, compression_spec); + CFH = InitCompressFileHandle(compression_spec); - if (!AH->OF) + if (CFH->open_func(filename, fn, mode, CFH)) { if (filename) pg_fatal("could not open output file \"%s\": %m", filename); else pg_fatal("could not open output file: %m"); } + + AH->OF = CFH; } -static cfp * +static CompressFileHandle * SaveOutput(ArchiveHandle *AH) { - return (cfp *) AH->OF; + return (CompressFileHandle *) AH->OF; } static void -RestoreOutput(ArchiveHandle *AH, cfp *savedOutput) +RestoreOutput(ArchiveHandle *AH, CompressFileHandle *savedOutput) { int res; errno = 0; - res = cfclose(AH->OF); + res = DestroyCompressFileHandle(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); @@ -1688,7 +1689,11 @@ ahwrite(const void *ptr, size_t size, size_t nmemb, ArchiveHandle *AH) else if (RestoringToDB(AH)) bytes_written = ExecuteSqlCommandBuf(&AH->public, (const char *) ptr, size * nmemb); else - bytes_written = cfwrite(ptr, size * nmemb, AH->OF); + { + CompressFileHandle *CFH = (CompressFileHandle *) AH->OF; + + bytes_written = CFH->write_func(ptr, size * nmemb, CFH); + } if (bytes_written != size * nmemb) WRITE_ERROR_EXIT; @@ -2030,6 +2035,18 @@ ReadStr(ArchiveHandle *AH) return buf; } +static bool +_fileExistsInDirectory(const char *dir, const char *filename) +{ + struct stat st; + char buf[MAXPGPATH]; + + if (snprintf(buf, MAXPGPATH, "%s/%s", dir, filename) >= MAXPGPATH) + pg_fatal("directory name too long: \"%s\"", dir); + + return (stat(buf, &st) == 0 && S_ISREG(st.st_mode)); +} + static int _discoverArchiveFormat(ArchiveHandle *AH) { @@ -2060,26 +2077,12 @@ _discoverArchiveFormat(ArchiveHandle *AH) */ if (stat(AH->fSpec, &st) == 0 && S_ISDIR(st.st_mode)) { - char buf[MAXPGPATH]; - - if (snprintf(buf, MAXPGPATH, "%s/toc.dat", AH->fSpec) >= MAXPGPATH) - pg_fatal("directory name too long: \"%s\"", - AH->fSpec); - if (stat(buf, &st) == 0 && S_ISREG(st.st_mode)) - { - AH->format = archDirectory; + AH->format = archDirectory; + if (_fileExistsInDirectory(AH->fSpec, "toc.dat")) return AH->format; - } - #ifdef HAVE_LIBZ - if (snprintf(buf, MAXPGPATH, "%s/toc.dat.gz", AH->fSpec) >= MAXPGPATH) - pg_fatal("directory name too long: \"%s\"", - AH->fSpec); - if (stat(buf, &st) == 0 && S_ISREG(st.st_mode)) - { - AH->format = archDirectory; + if (_fileExistsInDirectory(AH->fSpec, "toc.dat.gz")) return AH->format; - } #endif pg_fatal("directory \"%s\" does not appear to be a valid archive (\"toc.dat\" does not exist)", AH->fSpec); @@ -2177,6 +2180,7 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, SetupWorkerPtrType setupWorkerPtr) { ArchiveHandle *AH; + CompressFileHandle *CFH; pg_compress_specification out_compress_spec = {0}; pg_log_debug("allocating AH for %s, format %d", @@ -2232,7 +2236,10 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, /* Open stdout with no compression for AH output handle */ out_compress_spec.algorithm = PG_COMPRESSION_NONE; - AH->OF = cfdopen(dup(fileno(stdout)), PG_BINARY_A, &out_compress_spec); + CFH = InitCompressFileHandle(&out_compress_spec); + if (CFH->open_func(NULL, fileno(stdout), PG_BINARY_A, CFH)) + pg_fatal("could not open stdout for appending: %m"); + AH->OF = CFH; /* * On Windows, we need to use binary mode to read/write non-text files, @@ -3645,12 +3652,7 @@ WriteHead(ArchiveHandle *AH) AH->WriteBytePtr(AH, AH->intSize); AH->WriteBytePtr(AH, AH->offSize); AH->WriteBytePtr(AH, AH->format); - /* - * For now the compression type is implied by the level. This will need - * to change once support for more compression algorithms is added, - * requiring a format bump. - */ - WriteInt(AH, AH->compression_spec.level); + AH->WriteBytePtr(AH, AH->compression_spec.algorithm); crtm = *localtime(&AH->createDate); WriteInt(AH, crtm.tm_sec); WriteInt(AH, crtm.tm_min); @@ -3721,10 +3723,11 @@ ReadHead(ArchiveHandle *AH) pg_fatal("expected format (%d) differs from format found in file (%d)", AH->format, fmt); - /* Guess the compression method based on the level */ - AH->compression_spec.algorithm = PG_COMPRESSION_NONE; - if (AH->version >= K_VERS_1_2) + if (AH->version >= K_VERS_1_15) + AH->compression_spec.algorithm = AH->ReadBytePtr(AH); + else if (AH->version >= K_VERS_1_2) { + /* Guess the compression method based on the level */ if (AH->version < K_VERS_1_4) AH->compression_spec.level = AH->ReadBytePtr(AH); else @@ -3736,10 +3739,17 @@ ReadHead(ArchiveHandle *AH) else AH->compression_spec.algorithm = PG_COMPRESSION_GZIP; + if (AH->compression_spec.algorithm != PG_COMPRESSION_NONE) + { + bool unsupported = false; + #ifndef HAVE_LIBZ - if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) - pg_fatal("archive is compressed, but this installation does not support compression"); + if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) + unsupported = true; #endif + if (unsupported) + pg_fatal("archive is compressed, but this installation does not support compression"); + } if (AH->version >= K_VERS_1_4) { diff --git a/src/bin/pg_dump/pg_backup_archiver.h b/src/bin/pg_dump/pg_backup_archiver.h index 4725e49747b..18b38c17abc 100644 --- a/src/bin/pg_dump/pg_backup_archiver.h +++ b/src/bin/pg_dump/pg_backup_archiver.h @@ -65,10 +65,13 @@ #define K_VERS_1_13 MAKE_ARCHIVE_VERSION(1, 13, 0) /* change search_path * behavior */ #define K_VERS_1_14 MAKE_ARCHIVE_VERSION(1, 14, 0) /* add tableam */ +#define K_VERS_1_15 MAKE_ARCHIVE_VERSION(1, 15, 0) /* add + * compression_algorithm + * in header */ /* Current archive version number (the format we can output) */ #define K_VERS_MAJOR 1 -#define K_VERS_MINOR 14 +#define K_VERS_MINOR 15 #define K_VERS_REV 0 #define K_VERS_SELF MAKE_ARCHIVE_VERSION(K_VERS_MAJOR, K_VERS_MINOR, K_VERS_REV) diff --git a/src/bin/pg_dump/pg_backup_custom.c b/src/bin/pg_dump/pg_backup_custom.c index 0e87444de85..40cd90b7325 100644 --- a/src/bin/pg_dump/pg_backup_custom.c +++ b/src/bin/pg_dump/pg_backup_custom.c @@ -298,7 +298,9 @@ _StartData(ArchiveHandle *AH, TocEntry *te) _WriteByte(AH, BLK_DATA); /* Block type */ WriteInt(AH, te->dumpId); /* For sanity check */ - ctx->cs = AllocateCompressor(&AH->compression_spec, _CustomWriteFunc); + ctx->cs = AllocateCompressor(&AH->compression_spec, + NULL, + _CustomWriteFunc); } /* @@ -317,15 +319,15 @@ _WriteData(ArchiveHandle *AH, const void *data, size_t dLen) CompressorState *cs = ctx->cs; if (dLen > 0) - /* WriteDataToArchive() internally throws write errors */ - WriteDataToArchive(AH, cs, data, dLen); + /* writeData() internally throws write errors */ + cs->writeData(AH, cs, data, dLen); } /* * Called by the archiver when a dumper's 'DataDumper' routine has * finished. * - * Optional. + * Mandatory. */ static void _EndData(ArchiveHandle *AH, TocEntry *te) @@ -333,6 +335,8 @@ _EndData(ArchiveHandle *AH, TocEntry *te) lclContext *ctx = (lclContext *) AH->formatData; EndCompressor(AH, ctx->cs); + ctx->cs = NULL; + /* Send the end marker */ WriteInt(AH, 0); } @@ -377,7 +381,9 @@ _StartLO(ArchiveHandle *AH, TocEntry *te, Oid oid) WriteInt(AH, oid); - ctx->cs = AllocateCompressor(&AH->compression_spec, _CustomWriteFunc); + ctx->cs = AllocateCompressor(&AH->compression_spec, + NULL, + _CustomWriteFunc); } /* @@ -566,7 +572,12 @@ _PrintTocData(ArchiveHandle *AH, TocEntry *te) static void _PrintData(ArchiveHandle *AH) { - ReadDataFromArchive(AH, &AH->compression_spec, _CustomReadFunc); + CompressorState *cs; + + cs = AllocateCompressor(&AH->compression_spec, + _CustomReadFunc, NULL); + cs->readData(AH, cs); + EndCompressor(AH, cs); } static void diff --git a/src/bin/pg_dump/pg_backup_directory.c b/src/bin/pg_dump/pg_backup_directory.c index ffb8a0e4dd7..2d4baf58c22 100644 --- a/src/bin/pg_dump/pg_backup_directory.c +++ b/src/bin/pg_dump/pg_backup_directory.c @@ -50,9 +50,8 @@ typedef struct */ char *directory; - cfp *dataFH; /* currently open data file */ - - cfp *LOsTocFH; /* file handle for blobs.toc */ + CompressFileHandle *dataFH; /* currently open data file */ + CompressFileHandle *LOsTocFH; /* file handle for blobs.toc */ ParallelState *pstate; /* for parallel backup / restore */ } lclContext; @@ -198,11 +197,11 @@ InitArchiveFmt_Directory(ArchiveHandle *AH) else { /* Read Mode */ char fname[MAXPGPATH]; - cfp *tocFH; + CompressFileHandle *tocFH; setFilePath(AH, fname, "toc.dat"); - tocFH = cfopen_read(fname, PG_BINARY_R); + tocFH = InitDiscoverCompressFileHandle(fname, PG_BINARY_R); if (tocFH == NULL) pg_fatal("could not open input file \"%s\": %m", fname); @@ -218,7 +217,7 @@ InitArchiveFmt_Directory(ArchiveHandle *AH) ReadToc(AH); /* Nothing else in the file, so close it again... */ - if (cfclose(tocFH) != 0) + if (DestroyCompressFileHandle(tocFH) != 0) pg_fatal("could not close TOC file: %m"); ctx->dataFH = NULL; } @@ -327,9 +326,9 @@ _StartData(ArchiveHandle *AH, TocEntry *te) setFilePath(AH, fname, tctx->filename); - ctx->dataFH = cfopen_write(fname, PG_BINARY_W, - &AH->compression_spec); - if (ctx->dataFH == NULL) + ctx->dataFH = InitCompressFileHandle(&AH->compression_spec); + + if (ctx->dataFH->open_write_func(fname, PG_BINARY_W, ctx->dataFH)) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -346,15 +345,16 @@ static void _WriteData(ArchiveHandle *AH, const void *data, size_t dLen) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; errno = 0; - if (dLen > 0 && cfwrite(data, dLen, ctx->dataFH) != dLen) + if (dLen > 0 && CFH->write_func(data, dLen, CFH) != dLen) { /* if write didn't set errno, assume problem is no disk space */ if (errno == 0) errno = ENOSPC; pg_fatal("could not write to output file: %s", - get_cfp_error(ctx->dataFH)); + CFH->get_error_func(CFH)); } } @@ -370,7 +370,7 @@ _EndData(ArchiveHandle *AH, TocEntry *te) lclContext *ctx = (lclContext *) AH->formatData; /* Close the file */ - if (cfclose(ctx->dataFH) != 0) + if (DestroyCompressFileHandle(ctx->dataFH) != 0) pg_fatal("could not close data file: %m"); ctx->dataFH = NULL; @@ -385,26 +385,25 @@ _PrintFileData(ArchiveHandle *AH, char *filename) size_t cnt; char *buf; size_t buflen; - cfp *cfp; + CompressFileHandle *CFH; if (!filename) return; - cfp = cfopen_read(filename, PG_BINARY_R); - - if (!cfp) + CFH = InitDiscoverCompressFileHandle(filename, PG_BINARY_R); + if (!CFH) pg_fatal("could not open input file \"%s\": %m", filename); buf = pg_malloc(ZLIB_OUT_SIZE); buflen = ZLIB_OUT_SIZE; - while ((cnt = cfread(buf, buflen, cfp))) + while ((cnt = CFH->read_func(buf, buflen, CFH))) { ahwrite(buf, 1, cnt, AH); } free(buf); - if (cfclose(cfp) != 0) + if (DestroyCompressFileHandle(CFH) != 0) pg_fatal("could not close data file \"%s\": %m", filename); } @@ -435,6 +434,7 @@ _LoadLOs(ArchiveHandle *AH) { Oid oid; lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH; char tocfname[MAXPGPATH]; char line[MAXPGPATH]; @@ -442,14 +442,14 @@ _LoadLOs(ArchiveHandle *AH) setFilePath(AH, tocfname, "blobs.toc"); - ctx->LOsTocFH = cfopen_read(tocfname, PG_BINARY_R); + CFH = ctx->LOsTocFH = InitDiscoverCompressFileHandle(tocfname, PG_BINARY_R); if (ctx->LOsTocFH == NULL) pg_fatal("could not open large object TOC file \"%s\" for input: %m", tocfname); /* Read the LOs TOC file line-by-line, and process each LO */ - while ((cfgets(ctx->LOsTocFH, line, MAXPGPATH)) != NULL) + while ((CFH->gets_func(line, MAXPGPATH, CFH)) != NULL) { char lofname[MAXPGPATH + 1]; char path[MAXPGPATH]; @@ -464,11 +464,11 @@ _LoadLOs(ArchiveHandle *AH) _PrintFileData(AH, path); EndRestoreLO(AH, oid); } - if (!cfeof(ctx->LOsTocFH)) + if (!CFH->eof_func(CFH)) pg_fatal("error reading large object TOC file \"%s\"", tocfname); - if (cfclose(ctx->LOsTocFH) != 0) + if (DestroyCompressFileHandle(ctx->LOsTocFH) != 0) pg_fatal("could not close large object TOC file \"%s\": %m", tocfname); @@ -488,15 +488,16 @@ _WriteByte(ArchiveHandle *AH, const int i) { unsigned char c = (unsigned char) i; lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; errno = 0; - if (cfwrite(&c, 1, ctx->dataFH) != 1) + if (CFH->write_func(&c, 1, CFH) != 1) { /* if write didn't set errno, assume problem is no disk space */ if (errno == 0) errno = ENOSPC; pg_fatal("could not write to output file: %s", - get_cfp_error(ctx->dataFH)); + CFH->get_error_func(CFH)); } return 1; @@ -512,8 +513,9 @@ static int _ReadByte(ArchiveHandle *AH) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; - return cfgetc(ctx->dataFH); + return CFH->getc_func(CFH); } /* @@ -524,15 +526,16 @@ static void _WriteBuf(ArchiveHandle *AH, const void *buf, size_t len) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; errno = 0; - if (cfwrite(buf, len, ctx->dataFH) != len) + if (CFH->write_func(buf, len, CFH) != len) { /* if write didn't set errno, assume problem is no disk space */ if (errno == 0) errno = ENOSPC; pg_fatal("could not write to output file: %s", - get_cfp_error(ctx->dataFH)); + CFH->get_error_func(CFH)); } } @@ -545,12 +548,13 @@ static void _ReadBuf(ArchiveHandle *AH, void *buf, size_t len) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; /* - * If there was an I/O error, we already exited in cfread(), so here we + * If there was an I/O error, we already exited in readF(), so here we * exit on short reads. */ - if (cfread(buf, len, ctx->dataFH) != len) + if (CFH->read_func(buf, len, CFH) != len) pg_fatal("could not read from input file: end of file"); } @@ -573,7 +577,7 @@ _CloseArchive(ArchiveHandle *AH) if (AH->mode == archModeWrite) { - cfp *tocFH; + CompressFileHandle *tocFH; pg_compress_specification compression_spec = {0}; char fname[MAXPGPATH]; @@ -584,8 +588,8 @@ _CloseArchive(ArchiveHandle *AH) /* The TOC is always created uncompressed */ compression_spec.algorithm = PG_COMPRESSION_NONE; - tocFH = cfopen_write(fname, PG_BINARY_W, &compression_spec); - if (tocFH == NULL) + tocFH = InitCompressFileHandle(&compression_spec); + if (tocFH->open_write_func(fname, PG_BINARY_W, tocFH)) pg_fatal("could not open output file \"%s\": %m", fname); ctx->dataFH = tocFH; @@ -598,7 +602,7 @@ _CloseArchive(ArchiveHandle *AH) WriteHead(AH); AH->format = archDirectory; WriteToc(AH); - if (cfclose(tocFH) != 0) + if (DestroyCompressFileHandle(tocFH) != 0) pg_fatal("could not close TOC file: %m"); WriteDataChunks(AH, ctx->pstate); @@ -649,8 +653,8 @@ _StartLOs(ArchiveHandle *AH, TocEntry *te) /* The LO TOC file is never compressed */ compression_spec.algorithm = PG_COMPRESSION_NONE; - ctx->LOsTocFH = cfopen_write(fname, "ab", &compression_spec); - if (ctx->LOsTocFH == NULL) + ctx->LOsTocFH = InitCompressFileHandle(&compression_spec); + if (ctx->LOsTocFH->open_write_func(fname, "ab", ctx->LOsTocFH)) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -667,9 +671,8 @@ _StartLO(ArchiveHandle *AH, TocEntry *te, Oid oid) snprintf(fname, MAXPGPATH, "%s/blob_%u.dat", ctx->directory, oid); - ctx->dataFH = cfopen_write(fname, PG_BINARY_W, &AH->compression_spec); - - if (ctx->dataFH == NULL) + ctx->dataFH = InitCompressFileHandle(&AH->compression_spec); + if (ctx->dataFH->open_write_func(fname, PG_BINARY_W, ctx->dataFH)) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -682,18 +685,19 @@ static void _EndLO(ArchiveHandle *AH, TocEntry *te, Oid oid) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->LOsTocFH; char buf[50]; int len; - /* Close the LO data file itself */ - if (cfclose(ctx->dataFH) != 0) - pg_fatal("could not close LO data file: %m"); + /* Close the BLOB data file itself */ + if (DestroyCompressFileHandle(ctx->dataFH) != 0) + pg_fatal("could not close blob data file: %m"); ctx->dataFH = NULL; /* register the LO in blobs.toc */ len = snprintf(buf, sizeof(buf), "%u blob_%u.dat\n", oid, oid); - if (cfwrite(buf, len, ctx->LOsTocFH) != len) - pg_fatal("could not write to LOs TOC file"); + if (CFH->write_func(buf, len, CFH) != len) + pg_fatal("could not write to blobs TOC file"); } /* @@ -706,8 +710,8 @@ _EndLOs(ArchiveHandle *AH, TocEntry *te) { lclContext *ctx = (lclContext *) AH->formatData; - if (cfclose(ctx->LOsTocFH) != 0) - pg_fatal("could not close LOs TOC file: %m"); + if (DestroyCompressFileHandle(ctx->LOsTocFH) != 0) + pg_fatal("could not close blobs TOC file: %m"); ctx->LOsTocFH = NULL; } diff --git a/src/bin/pg_dump/t/002_pg_dump.pl b/src/bin/pg_dump/t/002_pg_dump.pl index 2eeef2a4783..f3ba9263213 100644 --- a/src/bin/pg_dump/t/002_pg_dump.pl +++ b/src/bin/pg_dump/t/002_pg_dump.pl @@ -94,7 +94,7 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/compression_gzip_custom.dump", ], - expected => qr/Compression: 1/, + expected => qr/Compression: gzip/, name => 'data content is gzip-compressed' }, }, @@ -239,8 +239,8 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/defaults_custom_format.dump", ], expected => $supports_gzip ? - qr/Compression: -1/ : - qr/Compression: 0/, + qr/Compression: gzip/ : + qr/Compression: none/, name => 'data content is gzip-compressed by default if available', }, }, @@ -264,8 +264,8 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/defaults_dir_format", ], expected => $supports_gzip ? - qr/Compression: -1/ : - qr/Compression: 0/, + qr/Compression: gzip/ : + qr/Compression: none/, name => 'data content is gzip-compressed by default', }, glob_patterns => [ diff --git a/src/tools/pginclude/cpluspluscheck b/src/tools/pginclude/cpluspluscheck index fe432e2cccc..62f3e9a81d4 100755 --- a/src/tools/pginclude/cpluspluscheck +++ b/src/tools/pginclude/cpluspluscheck @@ -150,6 +150,7 @@ do # pg_dump is not C++-clean because it uses "public" and "namespace" # as field names, which is unfortunate but we won't change it now. + test "$f" = src/bin/pg_dump/compress_gzip.h && continue test "$f" = src/bin/pg_dump/compress_io.h && continue test "$f" = src/bin/pg_dump/parallel.h && continue test "$f" = src/bin/pg_dump/pg_backup_archiver.h && continue diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 23bafec5f79..840191d680b 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -428,6 +428,7 @@ CompiledExprState CompositeIOData CompositeTypeStmt CompoundAffixFlag +CompressFileHandle CompressionLocation CompressorState ComputeXidHorizonsResult @@ -1034,6 +1035,7 @@ GucStack GucStackState GucStringAssignHook GucStringCheckHook +GzipCompressorState HANDLE HASHACTION HASHBUCKET -- 2.25.1 [text/x-diff] 0005-f.patch (5.0K, ../../[email protected]/6-0005-f.patch) download | inline diff: From 84404669ee9473d4ef7c24127453b37143324450 Mon Sep 17 00:00:00 2001 From: Justin Pryzby <[email protected]> Date: Sat, 14 Jan 2023 10:31:47 -0600 Subject: [PATCH 5/8] f! --- src/bin/pg_dump/compress_gzip.c | 13 +---------- src/bin/pg_dump/compress_io.c | 40 ++++++++++----------------------- 2 files changed, 13 insertions(+), 40 deletions(-) diff --git a/src/bin/pg_dump/compress_gzip.c b/src/bin/pg_dump/compress_gzip.c index 37c841c5a9b..a021d414624 100644 --- a/src/bin/pg_dump/compress_gzip.c +++ b/src/bin/pg_dump/compress_gzip.c @@ -220,8 +220,6 @@ InitCompressorGzip(CompressorState *cs, const pg_compress_specification *compres cs->writeData = WriteDataToArchiveGzip; cs->end = EndCompressorGzip; - cs->compression_spec = *compression_spec; - gzipcs = (GzipCompressorState *) pg_malloc0(sizeof(GzipCompressorState)); cs->private_data = gzipcs; @@ -291,17 +289,10 @@ static int Gzip_close(CompressFileHandle *CFH) { gzFile gzfp = (gzFile) CFH->private_data; - int save_errno; - int ret; CFH->private_data = NULL; - ret = gzclose(gzfp); - - save_errno = errno; - errno = save_errno; - - return ret; + return gzclose(gzfp); } static int @@ -386,8 +377,6 @@ InitCompressGzip(CompressFileHandle *CFH, const pg_compress_specification *compr CFH->eof_func = Gzip_eof; CFH->get_error_func = Gzip_get_error; - CFH->compression_spec = *compression_spec; - CFH->private_data = NULL; } #else /* HAVE_LIBZ */ diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index c4ac4042794..7a1fd318c12 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -39,14 +39,13 @@ * * The compressed stream API is a wrapper around the C standard fopen() and * libz's gzopen() APIs and custom LZ4 calls which provide similar - * functionality. It allows you to use the same functions for compressed and - * uncompressed streams. cfopen_read() first tries to open the file with given - * name, and if it fails, it tries to open the same file with the .gz suffix, - * failing that it tries to open the same file with the .lz4 suffix. - * cfopen_write() opens a file for writing, an extra argument specifies the - * method to use should the file be compressed, and adds the appropriate - * suffix, .gz or .lz4, to the filename if so. This allows you to easily handle - * both compressed and uncompressed files. + * libz's gzopen() APIs. It allows you to use the same functions for + * compressed and uncompressed streams. cfopen_read() first tries to open + * the file with given name, and if it fails, it tries to open the same + * file with the .gz suffix. cfopen_write() opens a file for writing, an + * extra argument specifies if the file should be compressed, and adds the + * .gz suffix to the filename if so. This allows you to easily handle both + * compressed and uncompressed files. * * IDENTIFICATION * src/bin/pg_dump/compress_io.c @@ -110,8 +109,6 @@ InitCompressorNone(CompressorState *cs, cs->readData = ReadDataFromArchiveNone; cs->writeData = WriteDataToArchiveNone; cs->end = EndCompressorNone; - - cs->compression_spec = *compression_spec; } /* Public interface routines */ @@ -126,7 +123,8 @@ AllocateCompressor(const pg_compress_specification *compression_spec, cs = (CompressorState *) pg_malloc0(sizeof(CompressorState)); cs->readF = readF; cs->writeF = writeF; - cs->compression_spec = *compression_spec; + cs->compression_spec = *compression_spec; // XXX: should do this here rather than every compressor ? + // does it even need to be passed at all ? switch (compression_spec->algorithm) { @@ -340,10 +338,9 @@ InitCompressFileHandle(const pg_compress_specification *compression_spec) * Open a file for reading. 'path' is the file to open, and 'mode' should * be either "r" or "rb". * - * If the file at 'path' does not exist, we append the "{.gz,.lz4}" suffix (i - * 'path' doesn't already have it) and try again. So if you pass "foo" as - * 'path', this will open either "foo" or "foo.gz" or "foo.lz4", trying in that - * order. + * If the file at 'path' does not exist, we append the ".gz" suffix (if 'path' + * doesn't already have it) and try again. So if you pass "foo" as 'path', + * this will open either "foo" or "foo.gz". * * On failure, return NULL with an error code in errno. */ @@ -355,8 +352,6 @@ InitDiscoverCompressFileHandle(const char *path, const char *mode) char *fname; pg_compress_specification compression_spec = {0}; - // compression_spec.algorithm = PG_COMPRESSION_NONE; - Assert(strcmp(mode, "r") == 0 || strcmp(mode, "rb") == 0); fname = strdup(path); @@ -381,17 +376,6 @@ InitDiscoverCompressFileHandle(const char *path, const char *mode) if (exists) compression_spec.algorithm = PG_COMPRESSION_GZIP; } -#endif -#ifdef USE_LZ4 - if (!exists) - { - free_keep_errno(fname); - fname = psprintf("%s.lz4", path); - exists = (stat(fname, &st) == 0); - - if (exists) - compression_spec.algorithm = PG_COMPRESSION_LZ4; - } #endif } -- 2.25.1 [text/x-diff] 0006-Add-LZ4-compression-in-pg_-dump-restore.patch (28.8K, ../../[email protected]/7-0006-Add-LZ4-compression-in-pg_-dump-restore.patch) download | inline diff: From 701484d05cab87147843b21971581bd61a0e01db Mon Sep 17 00:00:00 2001 From: Georgios Kokolatos <[email protected]> Date: Wed, 21 Dec 2022 09:49:36 +0000 Subject: [PATCH 6/8] Add LZ4 compression in pg_{dump|restore} Within compress_lz4.{c,h} the streaming API and a file API compression is implemented.. The first one, is aimed at inlined use cases and thus simple lz4.h calls can be used directly. The second one is generating output, or is parsing input, which can be read/generated via the lz4 utility. Wherever the LZ4F api does not implement all the functionality corresponding to fread(), fwrite(), fgets(), fgetc(), feof(), and fclose(), it has been implemented localy. --- doc/src/sgml/ref/pg_dump.sgml | 13 +- src/bin/pg_dump/Makefile | 2 + src/bin/pg_dump/compress_io.c | 13 +- src/bin/pg_dump/compress_lz4.c | 618 +++++++++++++++++++++++++++ src/bin/pg_dump/compress_lz4.h | 22 + src/bin/pg_dump/meson.build | 8 +- src/bin/pg_dump/pg_backup_archiver.c | 14 +- src/bin/pg_dump/pg_dump.c | 5 +- src/bin/pg_dump/t/002_pg_dump.pl | 82 +++- src/tools/pginclude/cpluspluscheck | 1 + src/tools/pgindent/typedefs.list | 1 + 11 files changed, 758 insertions(+), 21 deletions(-) create mode 100644 src/bin/pg_dump/compress_lz4.c create mode 100644 src/bin/pg_dump/compress_lz4.h diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml index 2c938cd7e14..49d218905fb 100644 --- a/doc/src/sgml/ref/pg_dump.sgml +++ b/doc/src/sgml/ref/pg_dump.sgml @@ -330,9 +330,10 @@ PostgreSQL documentation machine-readable format that <application>pg_restore</application> can read. A directory format archive can be manipulated with standard Unix tools; for example, files in an uncompressed archive - can be compressed with the <application>gzip</application> tool. - This format is compressed by default and also supports parallel - dumps. + can be compressed with the <application>gzip</application> or + <application>lz4</application>tool. + This format is compressed by default using <literal>gzip</literal> + and also supports parallel dumps. </para> </listitem> </varlistentry> @@ -654,7 +655,7 @@ PostgreSQL documentation <para> Specify the compression method and/or the compression level to use. The compression method can be set to <literal>gzip</literal> or - <literal>none</literal> for no compression. + <literal>lz4</literal> or <literal>none</literal> for no compression. A compression detail string can optionally be specified. If the detail string is an integer, it specifies the compression level. Otherwise, it should be a comma-separated list of items, each of the @@ -675,8 +676,8 @@ PostgreSQL documentation individual table-data segments, and the default is to compress using <literal>gzip</literal> at a moderate level. For plain text output, setting a nonzero compression level causes the entire output file to be compressed, - as though it had been fed through <application>gzip</application>; but the default - is not to compress. + as though it had been fed through <application>gzip</application> or + <application>lz4</application>; but the default is not to compress. </para> <para> The tar archive format currently does not support compression at all. diff --git a/src/bin/pg_dump/Makefile b/src/bin/pg_dump/Makefile index 7a19f5d6172..a1401377ab9 100644 --- a/src/bin/pg_dump/Makefile +++ b/src/bin/pg_dump/Makefile @@ -17,6 +17,7 @@ top_builddir = ../../.. include $(top_builddir)/src/Makefile.global export GZIP_PROGRAM=$(GZIP) +export LZ4 export with_icu override CPPFLAGS := -I$(libpq_srcdir) $(CPPFLAGS) @@ -25,6 +26,7 @@ LDFLAGS_INTERNAL += -L$(top_builddir)/src/fe_utils -lpgfeutils $(libpq_pgport) OBJS = \ $(WIN32RES) \ compress_gzip.o \ + compress_lz4.o \ compress_io.o \ dumputils.o \ parallel.o \ diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index 7a1fd318c12..95b18843080 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -59,6 +59,7 @@ #include "compress_io.h" #include "compress_gzip.h" +#include "compress_lz4.h" #include "pg_backup_utils.h" #ifdef HAVE_LIBZ @@ -135,7 +136,7 @@ AllocateCompressor(const pg_compress_specification *compression_spec, InitCompressorGzip(cs, compression_spec); break; case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); + InitCompressorLZ4(cs, compression_spec); break; case PG_COMPRESSION_ZSTD: pg_fatal("compression with %s is not yet supported", "ZSTD"); @@ -324,7 +325,7 @@ InitCompressFileHandle(const pg_compress_specification *compression_spec) InitCompressGzip(CFH, compression_spec); break; case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); + InitCompressLZ4(CFH, compression_spec); break; case PG_COMPRESSION_ZSTD: pg_fatal("compression with %s is not yet supported", "ZSTD"); @@ -338,11 +339,13 @@ InitCompressFileHandle(const pg_compress_specification *compression_spec) * Open a file for reading. 'path' is the file to open, and 'mode' should * be either "r" or "rb". * - * If the file at 'path' does not exist, we append the ".gz" suffix (if 'path' - * doesn't already have it) and try again. So if you pass "foo" as 'path', - * this will open either "foo" or "foo.gz". + * If the file at 'path' does not exist, we append the "{.gz,.lz4}" suffix (if + * 'path' doesn't already have it) and try again. So if you pass "foo" as + * 'path', this will open either "foo" or "foo.gz" or "foo.lz4", trying in that + * order. * * On failure, return NULL with an error code in errno. + * */ CompressFileHandle * InitDiscoverCompressFileHandle(const char *path, const char *mode) diff --git a/src/bin/pg_dump/compress_lz4.c b/src/bin/pg_dump/compress_lz4.c new file mode 100644 index 00000000000..c97e16187a0 --- /dev/null +++ b/src/bin/pg_dump/compress_lz4.c @@ -0,0 +1,618 @@ +/*------------------------------------------------------------------------- + * + * compress_lz4.c + * Routines for archivers to write an uncompressed or compressed data + * stream. + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_lz4.c + * + *------------------------------------------------------------------------- + */ +#include "postgres_fe.h" +#include "pg_backup_utils.h" + +#include "compress_lz4.h" + +#ifdef USE_LZ4 +#include <lz4.h> +#include <lz4frame.h> + +#define LZ4_OUT_SIZE (4 * 1024) +#define LZ4_IN_SIZE (16 * 1024) + +/* + * LZ4F_HEADER_SIZE_MAX first appeared in v1.7.5 of the library. + * Redefine it for installations with a lesser version. + */ +#ifndef LZ4F_HEADER_SIZE_MAX +#define LZ4F_HEADER_SIZE_MAX 32 +#endif + +/*---------------------- + * Compressor API + *---------------------- + */ + +typedef struct LZ4CompressorState +{ + char *outbuf; + size_t outsize; +} LZ4CompressorState; + +/* Private routines that support LZ4 compressed data I/O */ +static void ReadDataFromArchiveLZ4(ArchiveHandle *AH, CompressorState *cs); +static void WriteDataToArchiveLZ4(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen); +static void EndCompressorLZ4(ArchiveHandle *AH, CompressorState *cs); + +static void +ReadDataFromArchiveLZ4(ArchiveHandle *AH, CompressorState *cs) +{ + LZ4_streamDecode_t lz4StreamDecode; + char *buf; + char *decbuf; + size_t buflen; + size_t cnt; + + buflen = LZ4_IN_SIZE; + buf = pg_malloc(buflen); + decbuf = pg_malloc(buflen); + + LZ4_setStreamDecode(&lz4StreamDecode, NULL, 0); + + while ((cnt = cs->readF(AH, &buf, &buflen))) + { + int decBytes = LZ4_decompress_safe_continue(&lz4StreamDecode, + buf, decbuf, + cnt, buflen); + + ahwrite(decbuf, 1, decBytes, AH); + } + + pg_free(buf); + pg_free(decbuf); +} + +static void +WriteDataToArchiveLZ4(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen) +{ + LZ4CompressorState *LZ4cs = (LZ4CompressorState *) cs->private_data; + size_t compressed; + size_t requiredsize = LZ4_compressBound(dLen); + + if (requiredsize > LZ4cs->outsize) + { + LZ4cs->outbuf = pg_realloc(LZ4cs->outbuf, requiredsize); + LZ4cs->outsize = requiredsize; + } + + compressed = LZ4_compress_default(data, LZ4cs->outbuf, + dLen, LZ4cs->outsize); + + if (compressed <= 0) + pg_fatal("failed to LZ4 compress data"); + + cs->writeF(AH, LZ4cs->outbuf, compressed); +} + +static void +EndCompressorLZ4(ArchiveHandle *AH, CompressorState *cs) +{ + LZ4CompressorState *LZ4cs; + + LZ4cs = (LZ4CompressorState *) cs->private_data; + if (LZ4cs) + { + pg_free(LZ4cs->outbuf); + pg_free(LZ4cs); + cs->private_data = NULL; + } +} + + +/* Public routines that support LZ4 compressed data I/O */ +void +InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec) +{ + cs->readData = ReadDataFromArchiveLZ4; + cs->writeData = WriteDataToArchiveLZ4; + cs->end = EndCompressorLZ4; + + cs->compression_spec = compression_spec; + + /* Will be lazy init'd */ + cs->private_data = pg_malloc0(sizeof(LZ4CompressorState)); +} + +/*---------------------- + * Compress File API + *---------------------- + */ + +/* + * State needed for LZ4 (de)compression using the CompressFileHandle API. + */ +typedef struct LZ4File +{ + FILE *fp; + + LZ4F_preferences_t prefs; + + LZ4F_compressionContext_t ctx; + LZ4F_decompressionContext_t dtx; + + bool inited; + bool compressing; + + size_t buflen; + char *buffer; + + size_t overflowalloclen; + size_t overflowlen; + char *overflowbuf; + + size_t errcode; +} LZ4File; + +/* + * LZ4 equivalent to feof() or gzeof(). The end of file + * is reached if there is no decompressed output in the + * overflow buffer and the end of the file is reached. + */ +static int +LZ4File_eof(CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + + return fs->overflowlen == 0 && feof(fs->fp); +} + +static const char * +LZ4File_get_error(CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + const char *errmsg; + + if (LZ4F_isError(fs->errcode)) + errmsg = LZ4F_getErrorName(fs->errcode); + else + errmsg = strerror(errno); + + return errmsg; +} + +/* + * Prepare an already alloc'ed LZ4File struct for subsequent calls. + * + * It creates the nessary contexts for the operations. When compressing, + * it additionally writes the LZ4 header in the output stream. + */ +static int +LZ4File_init(LZ4File * fs, int size, bool compressing) +{ + size_t status; + + if (fs->inited) + return 0; + + fs->compressing = compressing; + fs->inited = true; + + if (fs->compressing) + { + fs->buflen = LZ4F_compressBound(LZ4_IN_SIZE, &fs->prefs); + if (fs->buflen < LZ4F_HEADER_SIZE_MAX) + fs->buflen = LZ4F_HEADER_SIZE_MAX; + + status = LZ4F_createCompressionContext(&fs->ctx, LZ4F_VERSION); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return 1; + } + + fs->buffer = pg_malloc(fs->buflen); + status = LZ4F_compressBegin(fs->ctx, fs->buffer, fs->buflen, + &fs->prefs); + + if (LZ4F_isError(status)) + { + fs->errcode = status; + return 1; + } + + if (fwrite(fs->buffer, 1, status, fs->fp) != status) + { + errno = errno ? : ENOSPC; + return 1; + } + } + else + { + status = LZ4F_createDecompressionContext(&fs->dtx, LZ4F_VERSION); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return 1; + } + + fs->buflen = size > LZ4_OUT_SIZE ? size : LZ4_OUT_SIZE; + fs->buffer = pg_malloc(fs->buflen); + + fs->overflowalloclen = fs->buflen; + fs->overflowbuf = pg_malloc(fs->overflowalloclen); + fs->overflowlen = 0; + } + + return 0; +} + +/* + * Read already decompressed content from the overflow buffer into 'ptr' up to + * 'size' bytes, if available. If the eol_flag is set, then stop at the first + * occurance of the new line char prior to 'size' bytes. + * + * Any unread content in the overflow buffer, is moved to the beginning. + */ +static int +LZ4File_read_overflow(LZ4File * fs, void *ptr, int size, bool eol_flag) +{ + char *p; + int readlen = 0; + + if (fs->overflowlen == 0) + return 0; + + if (fs->overflowlen >= size) + readlen = size; + else + readlen = fs->overflowlen; + + if (eol_flag && (p = memchr(fs->overflowbuf, '\n', readlen))) + /* Include the line terminating char */ + readlen = p - fs->overflowbuf + 1; + + memcpy(ptr, fs->overflowbuf, readlen); + fs->overflowlen -= readlen; + + if (fs->overflowlen > 0) + memmove(fs->overflowbuf, fs->overflowbuf + readlen, fs->overflowlen); + + return readlen; +} + +/* + * The workhorse for reading decompressed content out of an LZ4 compressed + * stream. + * + * It will read up to 'ptrsize' decompressed content, or up to the new line char + * if found first when the eol_flag is set. It is possible that the decompressed + * output generated by reading any compressed input via the LZ4F API, exceeds + * 'ptrsize'. Any exceeding decompressed content is stored at an overflow + * buffer within LZ4File. Of course, when the function is called, it will first + * try to consume any decompressed content already present in the overflow + * buffer, before decompressing new content. + */ +static int +LZ4File_read_internal(LZ4File * fs, void *ptr, int ptrsize, bool eol_flag) +{ + size_t dsize = 0; + size_t rsize; + size_t size = ptrsize; + bool eol_found = false; + + void *readbuf; + + /* Lazy init */ + if (!fs->inited && LZ4File_init(fs, size, false /* decompressing */ )) + return -1; + + /* Verfiy that there is enough space in the outbuf */ + if (size > fs->buflen) + { + fs->buflen = size; + fs->buffer = pg_realloc(fs->buffer, size); + } + + /* use already decompressed content if available */ + dsize = LZ4File_read_overflow(fs, ptr, size, eol_flag); + if (dsize == size || (eol_flag && memchr(ptr, '\n', dsize))) + return dsize; + + readbuf = pg_malloc(size); + + do + { + char *rp; + char *rend; + + rsize = fread(readbuf, 1, size, fs->fp); + if (rsize < size && !feof(fs->fp)) + return -1; + + rp = (char *) readbuf; + rend = (char *) readbuf + rsize; + + while (rp < rend) + { + size_t status; + size_t outlen = fs->buflen; + size_t read_remain = rend - rp; + + memset(fs->buffer, 0, outlen); + status = LZ4F_decompress(fs->dtx, fs->buffer, &outlen, + rp, &read_remain, NULL); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return -1; + } + + rp += read_remain; + + /* + * fill in what space is available in ptr if the eol flag is set, + * either skip if one already found or fill up to EOL if present + * in the outbuf + */ + if (outlen > 0 && dsize < size && eol_found == false) + { + char *p; + size_t lib = (eol_flag == 0) ? size - dsize : size - 1 - dsize; + size_t len = outlen < lib ? outlen : lib; + + if (eol_flag == true && + (p = memchr(fs->buffer, '\n', outlen)) && + (size_t) (p - fs->buffer + 1) <= len) + { + len = p - fs->buffer + 1; + eol_found = true; + } + + memcpy((char *) ptr + dsize, fs->buffer, len); + dsize += len; + + /* move what did not fit, if any, at the begining of the buf */ + if (len < outlen) + memmove(fs->buffer, fs->buffer + len, outlen - len); + outlen -= len; + } + + /* if there is available output, save it */ + if (outlen > 0) + { + while (fs->overflowlen + outlen > fs->overflowalloclen) + { + fs->overflowalloclen *= 2; + fs->overflowbuf = pg_realloc(fs->overflowbuf, + fs->overflowalloclen); + } + + memcpy(fs->overflowbuf + fs->overflowlen, fs->buffer, outlen); + fs->overflowlen += outlen; + } + } + } while (rsize == size && dsize < size && eol_found == 0); + + pg_free(readbuf); + + return (int) dsize; +} + +/* + * Compress size bytes from ptr and write them to the stream. + */ +static size_t +LZ4File_write(const void *ptr, size_t size, CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + size_t status; + int remaining = size; + + if (!fs->inited && LZ4File_init(fs, size, true)) + return -1; + + while (remaining > 0) + { + int chunk = remaining < LZ4_IN_SIZE ? remaining : LZ4_IN_SIZE; + + remaining -= chunk; + + status = LZ4F_compressUpdate(fs->ctx, fs->buffer, fs->buflen, + ptr, chunk, NULL); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return -1; + } + + if (fwrite(fs->buffer, 1, status, fs->fp) != status) + { + errno = errno ? : ENOSPC; + return 1; + } + } + + return size; +} + +/* + * fread() equivalent implementation for LZ4 compressed files. + */ +static size_t +LZ4File_read(void *ptr, size_t size, CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + int ret; + + ret = LZ4File_read_internal(fs, ptr, size, false); + if (ret != size && !LZ4File_eof(CFH)) + pg_fatal("could not read from input file: %s", LZ4File_get_error(CFH)); + + return ret; +} + +/* + * fgetc() equivalent implementation for LZ4 compressed files. + */ +static int +LZ4File_getc(CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + unsigned char c; + + if (LZ4File_read_internal(fs, &c, 1, false) != 1) + { + if (!LZ4File_eof(CFH)) + pg_fatal("could not read from input file: %s", LZ4File_get_error(CFH)); + else + pg_fatal("could not read from input file: end of file"); + } + + return c; +} + +/* + * fgets() equivalent implementation for LZ4 compressed files. + */ +static char * +LZ4File_gets(char *ptr, int size, CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + size_t dsize; + + dsize = LZ4File_read_internal(fs, ptr, size, true); + if (dsize < 0) + pg_fatal("could not read from input file: %s", LZ4File_get_error(CFH)); + + /* Done reading */ + if (dsize == 0) + return NULL; + + return ptr; +} + +/* + * Finalize (de)compression of a stream. When compressing it will write any + * remaining content and/or generated footer from the LZ4 API. + */ +static int +LZ4File_close(CompressFileHandle *CFH) +{ + FILE *fp; + LZ4File *fs = (LZ4File *) CFH->private_data; + size_t status; + int ret; + + fp = fs->fp; + if (fs->inited) + { + if (fs->compressing) + { + status = LZ4F_compressEnd(fs->ctx, fs->buffer, fs->buflen, NULL); + if (LZ4F_isError(status)) + pg_fatal("failed to end compression: %s", + LZ4F_getErrorName(status)); + else if ((ret = fwrite(fs->buffer, 1, status, fs->fp)) != status) + { + errno = errno ? : ENOSPC; + WRITE_ERROR_EXIT; + } + + status = LZ4F_freeCompressionContext(fs->ctx); + if (LZ4F_isError(status)) + pg_fatal("failed to end compression: %s", + LZ4F_getErrorName(status)); + } + else + { + status = LZ4F_freeDecompressionContext(fs->dtx); + if (LZ4F_isError(status)) + pg_fatal("failed to end decompression: %s", + LZ4F_getErrorName(status)); + pg_free(fs->overflowbuf); + } + + pg_free(fs->buffer); + } + + pg_free(fs); + + return fclose(fp); +} + +static int +LZ4File_open(const char *path, int fd, const char *mode, + CompressFileHandle *CFH) +{ + FILE *fp; + LZ4File *lz4fp = (LZ4File *) CFH->private_data; + + if (fd >= 0) + fp = fdopen(fd, mode); + else + fp = fopen(path, mode); + if (fp == NULL) + { + lz4fp->errcode = errno; + return 1; + } + + lz4fp->fp = fp; + + return 0; +} + +static int +LZ4File_open_write(const char *path, const char *mode, CompressFileHandle *CFH) +{ + char *fname; + int ret; + + fname = psprintf("%s.lz4", path); + ret = CFH->open_func(fname, -1, mode, CFH); + pg_free(fname); + + return ret; +} + +void +InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +{ + LZ4File *lz4fp; + + CFH->open_func = LZ4File_open; + CFH->open_write_func = LZ4File_open_write; + CFH->read_func = LZ4File_read; + CFH->write_func = LZ4File_write; + CFH->gets_func = LZ4File_gets; + CFH->getc_func = LZ4File_getc; + CFH->eof_func = LZ4File_eof; + CFH->close_func = LZ4File_close; + CFH->get_error_func = LZ4File_get_error; + + CFH->compression_spec = compression_spec; + lz4fp = pg_malloc0(sizeof(*lz4fp)); + if (CFH->compression_spec.level >= 0) + lz4fp->prefs.compressionLevel = CFH->compression_spec.level; + + CFH->private_data = lz4fp; +} +#else /* USE_LZ4 */ +void +InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec) +{ + pg_fatal("this build does not support compression with %s", "LZ4"); +} + +void +InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +{ + pg_fatal("this build does not support compression with %s", "LZ4"); +} +#endif /* USE_LZ4 */ diff --git a/src/bin/pg_dump/compress_lz4.h b/src/bin/pg_dump/compress_lz4.h new file mode 100644 index 00000000000..74595db1b98 --- /dev/null +++ b/src/bin/pg_dump/compress_lz4.h @@ -0,0 +1,22 @@ +/*------------------------------------------------------------------------- + * + * compress_lz4.h + * Interface to compress_io.c routines + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_lz4.h + * + *------------------------------------------------------------------------- + */ +#ifndef _COMPRESS_LZ4_H_ +#define _COMPRESS_LZ4_H_ + +#include "compress_io.h" + +extern void InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec); +extern void InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec); + +#endif /* _COMPRESS_LZ4_H_ */ diff --git a/src/bin/pg_dump/meson.build b/src/bin/pg_dump/meson.build index aa2c91829c0..473d40d456f 100644 --- a/src/bin/pg_dump/meson.build +++ b/src/bin/pg_dump/meson.build @@ -3,6 +3,7 @@ pg_dump_common_sources = files( 'compress_io.c', 'compress_gzip.c', + 'compress_lz4.c', 'dumputils.c', 'parallel.c', 'pg_backup_archiver.c', @@ -17,7 +18,7 @@ pg_dump_common_sources = files( pg_dump_common = static_library('libpgdump_common', pg_dump_common_sources, c_pch: pch_postgres_fe_h, - dependencies: [frontend_code, libpq, zlib], + dependencies: [frontend_code, libpq, lz4, zlib], kwargs: internal_lib_args, ) @@ -85,7 +86,10 @@ tests += { 'sd': meson.current_source_dir(), 'bd': meson.current_build_dir(), 'tap': { - 'env': {'GZIP_PROGRAM': gzip.path()}, + 'env': { + 'GZIP_PROGRAM': gzip.path(), + 'LZ4': program_lz4.found() ? program_lz4.path() : '', + }, 'tests': [ 't/001_basic.pl', 't/002_pg_dump.pl', diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index 9e9f1d626b5..40e96c93ebc 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -394,6 +394,10 @@ RestoreArchive(Archive *AHX) #ifndef HAVE_LIBZ if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) supports_compression = false; +#endif +#ifndef USE_LZ4 + if (AH->compression_spec.algorithm == PG_COMPRESSION_LZ4) + supports_compression = false; #endif if (supports_compression == false) pg_fatal("cannot restore from compressed archive (compression not supported in this installation)"); @@ -2073,7 +2077,7 @@ _discoverArchiveFormat(ArchiveHandle *AH) /* * Check if the specified archive is a directory. If so, check if - * there's a "toc.dat" (or "toc.dat.gz") file in it. + * there's a "toc.dat" (or "toc.dat.{gz,lz4}") file in it. */ if (stat(AH->fSpec, &st) == 0 && S_ISDIR(st.st_mode)) { @@ -2083,6 +2087,10 @@ _discoverArchiveFormat(ArchiveHandle *AH) #ifdef HAVE_LIBZ if (_fileExistsInDirectory(AH->fSpec, "toc.dat.gz")) return AH->format; +#endif +#ifdef USE_LZ4 + if (_fileExistsInDirectory(AH->fSpec, "toc.dat.lz4")) + return AH->format; #endif pg_fatal("directory \"%s\" does not appear to be a valid archive (\"toc.dat\" does not exist)", AH->fSpec); @@ -3746,6 +3754,10 @@ ReadHead(ArchiveHandle *AH) #ifndef HAVE_LIBZ if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) unsupported = true; +#endif +#ifndef USE_LZ4 + if (AH->compression_spec.algorithm == PG_COMPRESSION_LZ4) + unsupported = true; #endif if (unsupported) pg_fatal("archive is compressed, but this installation does not support compression"); diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c index 20f73729fac..224d2c900ce 100644 --- a/src/bin/pg_dump/pg_dump.c +++ b/src/bin/pg_dump/pg_dump.c @@ -715,13 +715,12 @@ main(int argc, char **argv) case PG_COMPRESSION_NONE: /* fallthrough */ case PG_COMPRESSION_GZIP: + /* fallthrough */ + case PG_COMPRESSION_LZ4: break; case PG_COMPRESSION_ZSTD: pg_fatal("compression with %s is not yet supported", "ZSTD"); break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; } /* diff --git a/src/bin/pg_dump/t/002_pg_dump.pl b/src/bin/pg_dump/t/002_pg_dump.pl index f3ba9263213..f497ec60407 100644 --- a/src/bin/pg_dump/t/002_pg_dump.pl +++ b/src/bin/pg_dump/t/002_pg_dump.pl @@ -139,6 +139,80 @@ my %pgdump_runs = ( args => [ '-d', "$tempdir/compression_gzip_plain.sql.gz", ], }, }, + + # Do not use --no-sync to give test coverage for data sync. + compression_lz4_custom => { + test_key => 'compression', + compile_option => 'lz4', + dump_cmd => [ + 'pg_dump', '--format=custom', + '--compress=lz4', "--file=$tempdir/compression_lz4_custom.dump", + 'postgres', + ], + restore_cmd => [ + 'pg_restore', + "--file=$tempdir/compression_lz4_custom.sql", + "$tempdir/compression_lz4_custom.dump", + ], + command_like => { + command => [ + 'pg_restore', + '-l', "$tempdir/compression_lz4_custom.dump", + ], + expected => qr/Compression: lz4/, + name => 'data content is lz4 compressed' + }, + }, + + # Do not use --no-sync to give test coverage for data sync. + compression_lz4_dir => { + test_key => 'compression', + compile_option => 'lz4', + dump_cmd => [ + 'pg_dump', '--jobs=2', + '--format=directory', '--compress=lz4:1', + "--file=$tempdir/compression_lz4_dir", 'postgres', + ], + # Give coverage for manually compressed blob.toc files during + # restore. + compress_cmd => { + program => $ENV{'LZ4'}, + args => [ + '-z', '-f', '--rm', + "$tempdir/compression_lz4_dir/blobs.toc", + "$tempdir/compression_lz4_dir/blobs.toc.lz4", + ], + }, + # Verify that data files where compressed + glob_patterns => [ + "$tempdir/compression_lz4_dir/toc.dat", + "$tempdir/compression_lz4_dir/*.dat.lz4", + ], + restore_cmd => [ + 'pg_restore', '--jobs=2', + "--file=$tempdir/compression_lz4_dir.sql", + "$tempdir/compression_lz4_dir", + ], + }, + + compression_lz4_plain => { + test_key => 'compression', + compile_option => 'lz4', + dump_cmd => [ + 'pg_dump', '--format=plain', '--compress=lz4', + "--file=$tempdir/compression_lz4_plain.sql.lz4", 'postgres', + ], + # Decompress the generated file to run through the tests. + compress_cmd => { + program => $ENV{'LZ4'}, + args => [ + '-d', '-f', + "$tempdir/compression_lz4_plain.sql.lz4", + "$tempdir/compression_lz4_plain.sql", + ], + }, + }, + clean => { dump_cmd => [ 'pg_dump', @@ -4175,11 +4249,11 @@ foreach my $run (sort keys %pgdump_runs) my $run_db = 'postgres'; # Skip command-level tests for gzip if there is no support for it. - if ( defined($pgdump_runs{$run}->{compile_option}) - && $pgdump_runs{$run}->{compile_option} eq 'gzip' - && !$supports_gzip) + if ($pgdump_runs{$run}->{compile_option} && + ($pgdump_runs{$run}->{compile_option} eq 'gzip' && !$supports_gzip) || + ($pgdump_runs{$run}->{compile_option} eq 'lz4' && !$supports_lz4)) { - note "$run: skipped due to no gzip support"; + note "$run: skipped due to no $pgdump_runs{$run}->{compile_option} support"; next; } diff --git a/src/tools/pginclude/cpluspluscheck b/src/tools/pginclude/cpluspluscheck index 62f3e9a81d4..2b461e797c6 100755 --- a/src/tools/pginclude/cpluspluscheck +++ b/src/tools/pginclude/cpluspluscheck @@ -152,6 +152,7 @@ do # as field names, which is unfortunate but we won't change it now. test "$f" = src/bin/pg_dump/compress_gzip.h && continue test "$f" = src/bin/pg_dump/compress_io.h && continue + test "$f" = src/bin/pg_dump/compress_lz4.h && continue test "$f" = src/bin/pg_dump/parallel.h && continue test "$f" = src/bin/pg_dump/pg_backup_archiver.h && continue test "$f" = src/bin/pg_dump/pg_dump.h && continue diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 840191d680b..232228d427c 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -1381,6 +1381,7 @@ LWLock LWLockHandle LWLockMode LWLockPadded +LZ4CompressorState LZ4F_compressionContext_t LZ4F_decompressOptions_t LZ4F_decompressionContext_t -- 2.25.1 [text/x-diff] 0007-f.patch (12.1K, ../../[email protected]/8-0007-f.patch) download | inline diff: From bc66e2b035ebd24de71ea41c909c62ca1aae2e2d Mon Sep 17 00:00:00 2001 From: Justin Pryzby <[email protected]> Date: Sat, 14 Jan 2023 10:45:02 -0600 Subject: [PATCH 7/8] f! --- src/bin/pg_dump/compress_io.c | 42 ++++++++++++++------------- src/bin/pg_dump/compress_io.h | 3 +- src/bin/pg_dump/compress_lz4.c | 34 +++++++++++----------- src/bin/pg_dump/compress_lz4.h | 4 +-- src/bin/pg_dump/pg_backup_directory.c | 10 +++++-- src/bin/pg_dump/t/002_pg_dump.pl | 2 +- 6 files changed, 51 insertions(+), 44 deletions(-) diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index 95b18843080..01bf3df0db0 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -39,13 +39,14 @@ * * The compressed stream API is a wrapper around the C standard fopen() and * libz's gzopen() APIs and custom LZ4 calls which provide similar - * libz's gzopen() APIs. It allows you to use the same functions for - * compressed and uncompressed streams. cfopen_read() first tries to open - * the file with given name, and if it fails, it tries to open the same - * file with the .gz suffix. cfopen_write() opens a file for writing, an - * extra argument specifies if the file should be compressed, and adds the - * .gz suffix to the filename if so. This allows you to easily handle both - * compressed and uncompressed files. + * functionality. It allows you to use the same functions for compressed and + * uncompressed streams. cfopen_read() first tries to open the file with given + * name, and if it fails, it tries to open the same file with the .gz suffix, + * failing that it tries to open the same file with the .lz4 suffix. + * cfopen_write() opens a file for writing, an extra argument specifies the + * method to use should the file be compressed, and adds the appropriate + * suffix, .gz or .lz4, to the filename if so. This allows you to easily handle + * both compressed and uncompressed files. * * IDENTIFICATION * src/bin/pg_dump/compress_io.c @@ -335,6 +336,14 @@ InitCompressFileHandle(const pg_compress_specification *compression_spec) return CFH; } +static bool +check_compressed_file(const char *path, char **fname, char *ext) +{ + free_keep_errno(*fname); + *fname = psprintf("%s.%s", path, ext); + return (access(*fname, F_OK) == 0); +} + /* * Open a file for reading. 'path' is the file to open, and 'mode' should * be either "r" or "rb". @@ -348,7 +357,7 @@ InitCompressFileHandle(const pg_compress_specification *compression_spec) * */ CompressFileHandle * -InitDiscoverCompressFileHandle(const char *path, const char *mode) +InitDiscoverCompressFileHandle(const char *path, const char *mode, pg_compress_algorithm alg) { CompressFileHandle *CFH = NULL; struct stat st; @@ -366,20 +375,13 @@ InitDiscoverCompressFileHandle(const char *path, const char *mode) bool exists; exists = (stat(path, &st) == 0); - /* avoid unused warning if it is not build with compression */ + /* avoid unused warning if it is not built with compression */ if (exists) compression_spec.algorithm = PG_COMPRESSION_NONE; -#ifdef HAVE_LIBZ - if (!exists) - { - free_keep_errno(fname); - fname = psprintf("%s.gz", path); - exists = (stat(fname, &st) == 0); - - if (exists) - compression_spec.algorithm = PG_COMPRESSION_GZIP; - } -#endif + else if (alg == PG_COMPRESSION_GZIP && check_compressed_file(path, &fname, "gz")) + compression_spec.algorithm = alg; + else if (alg == PG_COMPRESSION_LZ4 && check_compressed_file(path, &fname, "lz4")) + compression_spec.algorithm = alg; } CFH = InitCompressFileHandle(&compression_spec); diff --git a/src/bin/pg_dump/compress_io.h b/src/bin/pg_dump/compress_io.h index afe6b22efaf..2600182c469 100644 --- a/src/bin/pg_dump/compress_io.h +++ b/src/bin/pg_dump/compress_io.h @@ -93,7 +93,8 @@ extern CompressFileHandle *InitCompressFileHandle( const pg_compress_specification *compression_spec); extern CompressFileHandle *InitDiscoverCompressFileHandle(const char *path, - const char *mode); + const char *mode, + pg_compress_algorithm alg); extern int DestroyCompressFileHandle(CompressFileHandle *CFH); #endif diff --git a/src/bin/pg_dump/compress_lz4.c b/src/bin/pg_dump/compress_lz4.c index c97e16187a0..0e259a6251a 100644 --- a/src/bin/pg_dump/compress_lz4.c +++ b/src/bin/pg_dump/compress_lz4.c @@ -117,13 +117,13 @@ EndCompressorLZ4(ArchiveHandle *AH, CompressorState *cs) /* Public routines that support LZ4 compressed data I/O */ void -InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec) +InitCompressorLZ4(CompressorState *cs, const pg_compress_specification *compression_spec) { cs->readData = ReadDataFromArchiveLZ4; cs->writeData = WriteDataToArchiveLZ4; cs->end = EndCompressorLZ4; - cs->compression_spec = compression_spec; + cs->compression_spec = *compression_spec; /* Will be lazy init'd */ cs->private_data = pg_malloc0(sizeof(LZ4CompressorState)); @@ -189,7 +189,7 @@ LZ4File_get_error(CompressFileHandle *CFH) /* * Prepare an already alloc'ed LZ4File struct for subsequent calls. * - * It creates the nessary contexts for the operations. When compressing, + * It creates the necessary contexts for the operations. When compressing, * it additionally writes the LZ4 header in the output stream. */ static int @@ -228,7 +228,7 @@ LZ4File_init(LZ4File * fs, int size, bool compressing) if (fwrite(fs->buffer, 1, status, fs->fp) != status) { - errno = errno ? : ENOSPC; + errno = errno ? errno : ENOSPC; return 1; } } @@ -255,7 +255,7 @@ LZ4File_init(LZ4File * fs, int size, bool compressing) /* * Read already decompressed content from the overflow buffer into 'ptr' up to * 'size' bytes, if available. If the eol_flag is set, then stop at the first - * occurance of the new line char prior to 'size' bytes. + * occurrence of the new line char prior to 'size' bytes. * * Any unread content in the overflow buffer, is moved to the beginning. */ @@ -309,10 +309,10 @@ LZ4File_read_internal(LZ4File * fs, void *ptr, int ptrsize, bool eol_flag) void *readbuf; /* Lazy init */ - if (!fs->inited && LZ4File_init(fs, size, false /* decompressing */ )) + if (LZ4File_init(fs, size, false /* decompressing */ )) return -1; - /* Verfiy that there is enough space in the outbuf */ + /* Verify that there is enough space in the outbuf */ if (size > fs->buflen) { fs->buflen = size; @@ -363,10 +363,10 @@ LZ4File_read_internal(LZ4File * fs, void *ptr, int ptrsize, bool eol_flag) if (outlen > 0 && dsize < size && eol_found == false) { char *p; - size_t lib = (eol_flag == 0) ? size - dsize : size - 1 - dsize; + size_t lib = eol_flag ? size - 1 - dsize : size - dsize ; size_t len = outlen < lib ? outlen : lib; - if (eol_flag == true && + if (eol_flag && (p = memchr(fs->buffer, '\n', outlen)) && (size_t) (p - fs->buffer + 1) <= len) { @@ -377,7 +377,7 @@ LZ4File_read_internal(LZ4File * fs, void *ptr, int ptrsize, bool eol_flag) memcpy((char *) ptr + dsize, fs->buffer, len); dsize += len; - /* move what did not fit, if any, at the begining of the buf */ + /* move what did not fit, if any, at the beginning of the buf */ if (len < outlen) memmove(fs->buffer, fs->buffer + len, outlen - len); outlen -= len; @@ -414,7 +414,7 @@ LZ4File_write(const void *ptr, size_t size, CompressFileHandle *CFH) size_t status; int remaining = size; - if (!fs->inited && LZ4File_init(fs, size, true)) + if (LZ4File_init(fs, size, true)) return -1; while (remaining > 0) @@ -433,7 +433,7 @@ LZ4File_write(const void *ptr, size_t size, CompressFileHandle *CFH) if (fwrite(fs->buffer, 1, status, fs->fp) != status) { - errno = errno ? : ENOSPC; + errno = errno ? errno : ENOSPC; return 1; } } @@ -520,7 +520,7 @@ LZ4File_close(CompressFileHandle *CFH) LZ4F_getErrorName(status)); else if ((ret = fwrite(fs->buffer, 1, status, fs->fp)) != status) { - errno = errno ? : ENOSPC; + errno = errno ? errno : ENOSPC; WRITE_ERROR_EXIT; } @@ -582,7 +582,7 @@ LZ4File_open_write(const char *path, const char *mode, CompressFileHandle *CFH) } void -InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification *compression_spec) { LZ4File *lz4fp; @@ -596,7 +596,7 @@ InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compres CFH->close_func = LZ4File_close; CFH->get_error_func = LZ4File_get_error; - CFH->compression_spec = compression_spec; + CFH->compression_spec = *compression_spec; lz4fp = pg_malloc0(sizeof(*lz4fp)); if (CFH->compression_spec.level >= 0) lz4fp->prefs.compressionLevel = CFH->compression_spec.level; @@ -605,13 +605,13 @@ InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compres } #else /* USE_LZ4 */ void -InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec) +InitCompressorLZ4(CompressorState *cs, const pg_compress_specification *compression_spec) { pg_fatal("this build does not support compression with %s", "LZ4"); } void -InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification *compression_spec) { pg_fatal("this build does not support compression with %s", "LZ4"); } diff --git a/src/bin/pg_dump/compress_lz4.h b/src/bin/pg_dump/compress_lz4.h index 74595db1b98..69a3d9c171f 100644 --- a/src/bin/pg_dump/compress_lz4.h +++ b/src/bin/pg_dump/compress_lz4.h @@ -16,7 +16,7 @@ #include "compress_io.h" -extern void InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec); -extern void InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec); +extern void InitCompressorLZ4(CompressorState *cs, const pg_compress_specification *compression_spec); +extern void InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification *compression_spec); #endif /* _COMPRESS_LZ4_H_ */ diff --git a/src/bin/pg_dump/pg_backup_directory.c b/src/bin/pg_dump/pg_backup_directory.c index 2d4baf58c22..4845fd9368c 100644 --- a/src/bin/pg_dump/pg_backup_directory.c +++ b/src/bin/pg_dump/pg_backup_directory.c @@ -201,7 +201,8 @@ InitArchiveFmt_Directory(ArchiveHandle *AH) setFilePath(AH, fname, "toc.dat"); - tocFH = InitDiscoverCompressFileHandle(fname, PG_BINARY_R); + tocFH = InitDiscoverCompressFileHandle(fname, PG_BINARY_R, + PG_COMPRESSION_NONE); if (tocFH == NULL) pg_fatal("could not open input file \"%s\": %m", fname); @@ -390,7 +391,8 @@ _PrintFileData(ArchiveHandle *AH, char *filename) if (!filename) return; - CFH = InitDiscoverCompressFileHandle(filename, PG_BINARY_R); + CFH = InitDiscoverCompressFileHandle(filename, PG_BINARY_R, + AH->compression_spec.algorithm); if (!CFH) pg_fatal("could not open input file \"%s\": %m", filename); @@ -442,7 +444,8 @@ _LoadLOs(ArchiveHandle *AH) setFilePath(AH, tocfname, "blobs.toc"); - CFH = ctx->LOsTocFH = InitDiscoverCompressFileHandle(tocfname, PG_BINARY_R); + CFH = ctx->LOsTocFH = InitDiscoverCompressFileHandle(tocfname, PG_BINARY_R, + AH->compression_spec.algorithm); if (ctx->LOsTocFH == NULL) pg_fatal("could not open large object TOC file \"%s\" for input: %m", @@ -783,6 +786,7 @@ _PrepParallelRestore(ArchiveHandle *AH) { /* It might be compressed */ strlcat(fname, ".gz", sizeof(fname)); + // XXX if (stat(fname, &st) == 0) te->dataLength = st.st_size; } diff --git a/src/bin/pg_dump/t/002_pg_dump.pl b/src/bin/pg_dump/t/002_pg_dump.pl index f497ec60407..263995a2b7a 100644 --- a/src/bin/pg_dump/t/002_pg_dump.pl +++ b/src/bin/pg_dump/t/002_pg_dump.pl @@ -183,7 +183,7 @@ my %pgdump_runs = ( "$tempdir/compression_lz4_dir/blobs.toc.lz4", ], }, - # Verify that data files where compressed + # Verify that data files were compressed glob_patterns => [ "$tempdir/compression_lz4_dir/toc.dat", "$tempdir/compression_lz4_dir/*.dat.lz4", -- 2.25.1 [text/x-diff] 0008-TMP-pg_dump-use-lz4-by-default-for-CI-only.patch (2.0K, ../../[email protected]/9-0008-TMP-pg_dump-use-lz4-by-default-for-CI-only.patch) download | inline diff: From 6c21acef733a52aa95fd517d73033d0fe9e5efbd Mon Sep 17 00:00:00 2001 From: Justin Pryzby <[email protected]> Date: Wed, 4 Jan 2023 21:21:53 -0600 Subject: [PATCH 8/8] TMP: pg_dump: use lz4 by default, for CI only --- src/bin/pg_dump/pg_dump.c | 7 +++++-- src/bin/pg_dump/t/002_pg_dump.pl | 8 ++++---- 2 files changed, 9 insertions(+), 6 deletions(-) diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c index 224d2c900ce..cf5083c432f 100644 --- a/src/bin/pg_dump/pg_dump.c +++ b/src/bin/pg_dump/pg_dump.c @@ -733,8 +733,11 @@ main(int argc, char **argv) #ifdef HAVE_LIBZ parse_compress_specification(PG_COMPRESSION_GZIP, NULL, &compression_spec); -#else - /* Nothing to do in the default case */ +#endif + +#ifdef USE_LZ4 + parse_compress_specification(PG_COMPRESSION_LZ4, NULL, + &compression_spec); #endif } diff --git a/src/bin/pg_dump/t/002_pg_dump.pl b/src/bin/pg_dump/t/002_pg_dump.pl index 263995a2b7a..3485ebca57d 100644 --- a/src/bin/pg_dump/t/002_pg_dump.pl +++ b/src/bin/pg_dump/t/002_pg_dump.pl @@ -313,9 +313,9 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/defaults_custom_format.dump", ], expected => $supports_gzip ? - qr/Compression: gzip/ : + qr/Compression: lz4/ : qr/Compression: none/, - name => 'data content is gzip-compressed by default if available', + name => 'data content is lz4-compressed by default if available', }, }, @@ -338,7 +338,7 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/defaults_dir_format", ], expected => $supports_gzip ? - qr/Compression: gzip/ : + qr/Compression: lz4/ : qr/Compression: none/, name => 'data content is gzip-compressed by default', }, @@ -346,7 +346,7 @@ my %pgdump_runs = ( "$tempdir/defaults_dir_format/toc.dat", "$tempdir/defaults_dir_format/blobs.toc", $supports_gzip ? - "$tempdir/defaults_dir_format/*.dat.gz" : + "$tempdir/defaults_dir_format/*.dat.lz4" : "$tempdir/defaults_dir_format/*.dat", ], }, -- 2.25.1 ^ permalink raw reply [nested|flat] 29+ messages in thread
* Re: Add LZ4 compression in pg_dump @ 2023-01-16 01:28 Michael Paquier <[email protected]> parent: Justin Pryzby <[email protected]> 1 sibling, 1 reply; 29+ messages in thread From: Michael Paquier @ 2023-01-16 01:28 UTC (permalink / raw) To: Justin Pryzby <[email protected]>; +Cc: [email protected]; [email protected]; Rachel Heaton <[email protected]> On Sat, Jan 14, 2023 at 03:43:09PM -0600, Justin Pryzby wrote: > On Sun, Jan 08, 2023 at 01:45:25PM -0600, Justin Pryzby wrote: >> pg_compress_specification is being passed by value, but I think it >> should be passed as a pointer, as is done everywhere else. > > ISTM that was an issue with 5e73a6048, affecting a few public and > private functions. I wrote a pre-preparatory patch which changes to > pass by reference. The functions changed by 0001 are cfopen[_write](), AllocateCompressor() and ReadDataFromArchive(). Why is it a good idea to change these interfaces which basically exist to handle inputs? Is there some benefit in changing compression_spec within the internals of these routines before going back one layer down to their callers? Changing the compression_spec on-the-fly in these internal paths could be risky, actually, no? > And addressed a handful of other issues I reported as separate fixup > commits. And changed to use LZ4 by default for CI. Are your slight changes shaped as of 0003-f.patch, 0005-f.patch and 0007-f.patch on top of the original patches sent by Georgios? > I also rebased my 2 year old patch to support zstd in pg_dump. I hope > it can finally added for v16. I'll send it for the next CF if these > patches progress. Good idea to see if what you have done for zstd fits with what's presented here. >> pg_compress_algorithm is being writen directly into the pg_dump header. Do you mean that this is what happens once the patch series 0001~0008 sent upthread is applied on HEAD? >> Currently, I think that's not an externally-visible value (it could be >> renumbered, theoretically even in a minor release). Maybe there should >> be a "private" enum for encoding the pg_dump header, similar to >> WAL_COMPRESSION_LZ4 vs BKPIMAGE_COMPRESS_LZ4 ? Or else a comment there >> should warn that the values are encoded in pg_dump, and must never be >> changed. > > Michael, WDYT ? Changing the order of the members in an enum would cause an ABI breakage, so that would not happen, and we tend to be very careful about that. Appending new members would be fine, though. FWIW, I'd rather avoid adding more enums that would just be exact maps to pg_compress_algorithm. - /* - * For now the compression type is implied by the level. This will need - * to change once support for more compression algorithms is added, - * requiring a format bump. - */ - WriteInt(AH, AH->compression_spec.level); + AH->WriteBytePtr(AH, AH->compression_spec.algorithm); I may be missing something here, but it seems to me that you ought to store as well the level in the dump header, or it would not be possible to report in the dump's description what was used? Hence, K_VERS_1_15 should imply that we have both the method compression and the compression level. -- Michael Attachments: [application/pgp-signature] signature.asc (833B, ../../[email protected]/2-signature.asc) download ^ permalink raw reply [nested|flat] 29+ messages in thread
* Re: Add LZ4 compression in pg_dump @ 2023-01-16 01:56 Justin Pryzby <[email protected]> parent: Michael Paquier <[email protected]> 0 siblings, 1 reply; 29+ messages in thread From: Justin Pryzby @ 2023-01-16 01:56 UTC (permalink / raw) To: Michael Paquier <[email protected]>; +Cc: [email protected]; [email protected]; Rachel Heaton <[email protected]> On Mon, Jan 16, 2023 at 10:28:50AM +0900, Michael Paquier wrote: > On Sat, Jan 14, 2023 at 03:43:09PM -0600, Justin Pryzby wrote: > > On Sun, Jan 08, 2023 at 01:45:25PM -0600, Justin Pryzby wrote: > >> pg_compress_specification is being passed by value, but I think it > >> should be passed as a pointer, as is done everywhere else. > > > > ISTM that was an issue with 5e73a6048, affecting a few public and > > private functions. I wrote a pre-preparatory patch which changes to > > pass by reference. > > The functions changed by 0001 are cfopen[_write](), > AllocateCompressor() and ReadDataFromArchive(). Why is it a good idea > to change these interfaces which basically exist to handle inputs? I changed to pass pg_compress_specification as a pointer, since that's the usual convention for structs, as followed by the existing uses of pg_compress_specification. > Is there some benefit in changing compression_spec within the > internals of these routines before going back one layer down to their > callers? Changing the compression_spec on-the-fly in these internal > paths could be risky, actually, no? I think what you're saying is that if the spec is passed as a pointer, then the called functions shouldn't set spec->algorithm=something. I agree that if they need to do that, they should use a local variable. Which looks to be true for the functions that were changed in 001. > > And addressed a handful of other issues I reported as separate fixup > > commits. And changed to use LZ4 by default for CI. > > Are your slight changes shaped as of 0003-f.patch, 0005-f.patch and > 0007-f.patch on top of the original patches sent by Georgios? Yes, the original patches, rebased as needed on top of HEAD and 001... > >> pg_compress_algorithm is being writen directly into the pg_dump header. > > Do you mean that this is what happens once the patch series 0001~0008 > sent upthread is applied on HEAD? Yes > - /* > - * For now the compression type is implied by the level. This will need > - * to change once support for more compression algorithms is added, > - * requiring a format bump. > - */ > - WriteInt(AH, AH->compression_spec.level); > + AH->WriteBytePtr(AH, AH->compression_spec.algorithm); > > I may be missing something here, but it seems to me that you ought to > store as well the level in the dump header, or it would not be > possible to report in the dump's description what was used? Hence, > K_VERS_1_15 should imply that we have both the method compression and > the compression level. Maybe. But the "level" isn't needed for decompression for any case I'm aware of. Also, dumps with the default compression level currently say: "Compression: -1", which does't seem valuable. -- Justin ^ permalink raw reply [nested|flat] 29+ messages in thread
* Re: Add LZ4 compression in pg_dump @ 2023-01-16 02:54 Michael Paquier <[email protected]> parent: Justin Pryzby <[email protected]> 0 siblings, 1 reply; 29+ messages in thread From: Michael Paquier @ 2023-01-16 02:54 UTC (permalink / raw) To: Justin Pryzby <[email protected]>; +Cc: [email protected]; [email protected]; Rachel Heaton <[email protected]> On Sun, Jan 15, 2023 at 07:56:25PM -0600, Justin Pryzby wrote: > On Mon, Jan 16, 2023 at 10:28:50AM +0900, Michael Paquier wrote: >> The functions changed by 0001 are cfopen[_write](), >> AllocateCompressor() and ReadDataFromArchive(). Why is it a good idea >> to change these interfaces which basically exist to handle inputs? > > I changed to pass pg_compress_specification as a pointer, since that's > the usual convention for structs, as followed by the existing uses of > pg_compress_specification. Okay, but what do we gain here? It seems to me that this introduces the risk that a careless change in one of the internal routines if they change slight;ly compress_spec, hence impacting any of their callers? Or is that fixing an actual bug (except if I am missing your point, that does not seem to be the case)? >> Is there some benefit in changing compression_spec within the >> internals of these routines before going back one layer down to their >> callers? Changing the compression_spec on-the-fly in these internal >> paths could be risky, actually, no? > > I think what you're saying is that if the spec is passed as a pointer, > then the called functions shouldn't set spec->algorithm=something. Yes. HEAD makes sure of that, 0001 would not prevent that. So I am a bit confused in seeing how this is a benefit. -- Michael Attachments: [application/pgp-signature] signature.asc (833B, ../../[email protected]/2-signature.asc) download ^ permalink raw reply [nested|flat] 29+ messages in thread
* Re: Add LZ4 compression in pg_dump @ 2023-01-16 15:14 [email protected] parent: Michael Paquier <[email protected]> 0 siblings, 0 replies; 29+ messages in thread From: [email protected] @ 2023-01-16 15:14 UTC (permalink / raw) To: Michael Paquier <[email protected]>; +Cc: Justin Pryzby <[email protected]>; [email protected]; Rachel Heaton <[email protected]> Hi, I admit I am completely at lost as to what is expected from me anymore. I had posted v19-0001 for a committer's consideration and v19-000{2,3} for completeness. Please find a rebased v20 attached. Also please let me know if I should silently step away from it and let other people lead it. I would be glad to comply either way. Cheers, //Georgios ------- Original Message ------- On Monday, January 16th, 2023 at 3:54 AM, Michael Paquier <[email protected]> wrote: > > > On Sun, Jan 15, 2023 at 07:56:25PM -0600, Justin Pryzby wrote: > > > On Mon, Jan 16, 2023 at 10:28:50AM +0900, Michael Paquier wrote: > > > > > The functions changed by 0001 are cfopen_write, > > > AllocateCompressor() and ReadDataFromArchive(). Why is it a good idea > > > to change these interfaces which basically exist to handle inputs? > > > > I changed to pass pg_compress_specification as a pointer, since that's > > the usual convention for structs, as followed by the existing uses of > > pg_compress_specification. > > > Okay, but what do we gain here? It seems to me that this introduces > the risk that a careless change in one of the internal routines if > they change slight;ly compress_spec, hence impacting any of their > callers? Or is that fixing an actual bug (except if I am missing your > point, that does not seem to be the case)? > > > > Is there some benefit in changing compression_spec within the > > > internals of these routines before going back one layer down to their > > > callers? Changing the compression_spec on-the-fly in these internal > > > paths could be risky, actually, no? > > > > I think what you're saying is that if the spec is passed as a pointer, > > then the called functions shouldn't set spec->algorithm=something. > > > Yes. HEAD makes sure of that, 0001 would not prevent that. So I am a > bit confused in seeing how this is a benefit. > -- > Michael Attachments: [text/x-patch] v20-0001-Prepare-pg_dump-internals-for-additional-compres.patch (21.9K, ../../twKgLFeqHEshCjdrbnzkwTMHiqfqvrrPSni7eKZjcxkForO-sI7HL4AKFgk9En1nGk3i4TXOY_An2xnCzv8MNjoLz4iupACaRB-0VRV20V4=@pm.me/2-v20-0001-Prepare-pg_dump-internals-for-additional-compres.patch) download | inline diff: From 348a306b47148790c602b2c208ff3345befa8eb7 Mon Sep 17 00:00:00 2001 From: Georgios Kokolatos <[email protected]> Date: Mon, 16 Jan 2023 14:56:31 +0000 Subject: [PATCH v20 1/3] Prepare pg_dump internals for additional compression methods. Commit bf9aa490db introduced cfp in compress_io.{c,h} with the intent of unifying compression related code and allow for the introduction of additional archive formats. However, pg_backup_archiver.c was not using that API. This commit teaches pg_backup_archiver.c about it and is using it throughout. --- src/bin/pg_dump/compress_io.c | 389 +++++++++++++++++++-------- src/bin/pg_dump/pg_backup_archiver.c | 128 +++------ src/bin/pg_dump/pg_backup_archiver.h | 27 +- 3 files changed, 318 insertions(+), 226 deletions(-) diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index 7a2c80bbc4..1db973b6a2 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -56,6 +56,10 @@ #include "compress_io.h" #include "pg_backup_utils.h" +#ifdef HAVE_LIBZ +#include <zlib.h> +#endif + /*---------------------- * Compressor API *---------------------- @@ -128,15 +132,24 @@ ReadDataFromArchive(ArchiveHandle *AH, const pg_compress_specification compression_spec, ReadFunc readF) { - if (compression_spec.algorithm == PG_COMPRESSION_NONE) - ReadDataFromArchiveNone(AH, readF); - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) + switch (compression_spec.algorithm) { + case PG_COMPRESSION_NONE: + ReadDataFromArchiveNone(AH, readF); + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - ReadDataFromArchiveZlib(AH, readF); + ReadDataFromArchiveZlib(AH, readF); #else - pg_fatal("this build does not support compression with %s", "gzip"); + pg_fatal("this build does not support compression with %s", "gzip"); #endif + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } } @@ -149,6 +162,9 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, { switch (cs->compression_spec.algorithm) { + case PG_COMPRESSION_NONE: + WriteDataToArchiveNone(AH, cs, data, dLen); + break; case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ WriteDataToArchiveZlib(AH, cs, data, dLen); @@ -156,13 +172,11 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, pg_fatal("this build does not support compression with %s", "gzip"); #endif break; - case PG_COMPRESSION_NONE: - WriteDataToArchiveNone(AH, cs, data, dLen); - break; case PG_COMPRESSION_LZ4: - /* fallthrough */ + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; case PG_COMPRESSION_ZSTD: - pg_fatal("invalid compression method"); + pg_fatal("compression with %s is not yet supported", "ZSTD"); break; } } @@ -173,10 +187,26 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, void EndCompressor(ArchiveHandle *AH, CompressorState *cs) { + switch (cs->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (cs->compression_spec.algorithm == PG_COMPRESSION_GZIP) - EndCompressorZlib(AH, cs); + EndCompressorZlib(AH, cs); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + free(cs); } @@ -391,10 +421,8 @@ WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, */ struct cfp { - FILE *uncompressedfp; -#ifdef HAVE_LIBZ - gzFile compressedfp; -#endif + pg_compress_specification compression_spec; + void *fp; }; #ifdef HAVE_LIBZ @@ -490,127 +518,202 @@ cfopen_write(const char *path, const char *mode, } /* - * Opens file 'path' in 'mode'. If compression is GZIP, the file - * is opened with libz gzopen(), otherwise with plain fopen(). + * This is the workhorse for cfopen() or cfdopen(). It opens file 'path' or + * associates a stream 'fd', if 'fd' is a valid descriptor, in 'mode'. The + * descriptor is not dup'ed and it is the caller's responsibility to do so. + * The caller must verify that the 'compress_algorithm' is supported by the + * current build. * * On failure, return NULL with an error code in errno. */ -cfp * -cfopen(const char *path, const char *mode, - const pg_compress_specification compression_spec) +static cfp * +cfopen_internal(const char *path, int fd, const char *mode, + pg_compress_specification compression_spec) { cfp *fp = pg_malloc(sizeof(cfp)); - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) + fp->compression_spec = compression_spec; + + switch (compression_spec.algorithm) { -#ifdef HAVE_LIBZ - if (compression_spec.level != Z_DEFAULT_COMPRESSION) - { - /* user has specified a compression level, so tell zlib to use it */ - char mode_compression[32]; + case PG_COMPRESSION_NONE: + if (fd >= 0) + fp->fp = fdopen(fd, mode); + else + fp->fp = fopen(path, mode); + if (fp->fp == NULL) + { + free_keep_errno(fp); + fp = NULL; + } - snprintf(mode_compression, sizeof(mode_compression), "%s%d", - mode, compression_spec.level); - fp->compressedfp = gzopen(path, mode_compression); - } - else - { - /* don't specify a level, just use the zlib default */ - fp->compressedfp = gzopen(path, mode); - } + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + if (compression_spec.level != Z_DEFAULT_COMPRESSION) + { + /* + * user has specified a compression level, so tell zlib to use + * it + */ + char mode_compression[32]; + + snprintf(mode_compression, sizeof(mode_compression), "%s%d", + mode, compression_spec.level); + if (fd >= 0) + fp->fp = gzdopen(fd, mode_compression); + else + fp->fp = gzopen(path, mode_compression); + } + else + { + /* don't specify a level, just use the zlib default */ + if (fd >= 0) + fp->fp = gzdopen(fd, mode); + else + fp->fp = gzopen(path, mode); + } - fp->uncompressedfp = NULL; - if (fp->compressedfp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } + if (fp->fp == NULL) + { + free_keep_errno(fp); + fp = NULL; + } #else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - } - else - { -#ifdef HAVE_LIBZ - fp->compressedfp = NULL; + pg_fatal("this build does not support compression with %s", "gzip"); #endif - fp->uncompressedfp = fopen(path, mode); - if (fp->uncompressedfp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } return fp; } +cfp * +cfopen(const char *path, const char *mode, + const pg_compress_specification compression_spec) +{ + return cfopen_internal(path, -1, mode, compression_spec); +} + +cfp * +cfdopen(int fd, const char *mode, + const pg_compress_specification compression_spec) +{ + return cfopen_internal(NULL, fd, mode, compression_spec); +} int cfread(void *ptr, int size, cfp *fp) { - int ret; + int ret = 0; if (size == 0) return 0; -#ifdef HAVE_LIBZ - if (fp->compressedfp) + switch (fp->compression_spec.algorithm) { - ret = gzread(fp->compressedfp, ptr, size); - if (ret != size && !gzeof(fp->compressedfp)) - { - int errnum; - const char *errmsg = gzerror(fp->compressedfp, &errnum); + case PG_COMPRESSION_NONE: + ret = fread(ptr, 1, size, (FILE *) fp->fp); + if (ret != size && !feof((FILE *) fp->fp)) + READ_ERROR_EXIT((FILE *) fp->fp); - pg_fatal("could not read from input file: %s", - errnum == Z_ERRNO ? strerror(errno) : errmsg); - } - } - else + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + ret = gzread((gzFile) fp->fp, ptr, size); + if (ret != size && !gzeof((gzFile) fp->fp)) + { + int errnum; + const char *errmsg = gzerror((gzFile) fp->fp, &errnum); + + pg_fatal("could not read from input file: %s", + errnum == Z_ERRNO ? strerror(errno) : errmsg); + } +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - { - ret = fread(ptr, 1, size, fp->uncompressedfp); - if (ret != size && !feof(fp->uncompressedfp)) - READ_ERROR_EXIT(fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } + return ret; } int cfwrite(const void *ptr, int size, cfp *fp) { + int ret = 0; + + switch (fp->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + ret = fwrite(ptr, 1, size, (FILE *) fp->fp); + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (fp->compressedfp) - return gzwrite(fp->compressedfp, ptr, size); - else + ret = gzwrite((gzFile) fp->fp, ptr, size); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - return fwrite(ptr, 1, size, fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + + return ret; } int cfgetc(cfp *fp) { - int ret; + int ret = 0; -#ifdef HAVE_LIBZ - if (fp->compressedfp) + switch (fp->compression_spec.algorithm) { - ret = gzgetc(fp->compressedfp); - if (ret == EOF) - { - if (!gzeof(fp->compressedfp)) - pg_fatal("could not read from input file: %s", strerror(errno)); - else - pg_fatal("could not read from input file: end of file"); - } - } - else + case PG_COMPRESSION_NONE: + ret = fgetc((FILE *) fp->fp); + if (ret == EOF) + READ_ERROR_EXIT((FILE *) fp->fp); + + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + ret = gzgetc((gzFile) fp->fp); + if (ret == EOF) + { + if (!gzeof((gzFile) fp->fp)) + pg_fatal("could not read from input file: %s", strerror(errno)); + else + pg_fatal("could not read from input file: end of file"); + } +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - { - ret = fgetc(fp->uncompressedfp); - if (ret == EOF) - READ_ERROR_EXIT(fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } return ret; @@ -619,65 +722,119 @@ cfgetc(cfp *fp) char * cfgets(cfp *fp, char *buf, int len) { + char *ret = NULL; + + switch (fp->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + ret = fgets(buf, len, (FILE *) fp->fp); + + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (fp->compressedfp) - return gzgets(fp->compressedfp, buf, len); - else + ret = gzgets((gzFile) fp->fp, buf, len); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - return fgets(buf, len, fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + + return ret; } int cfclose(cfp *fp) { - int result; + int ret = 0; if (fp == NULL) { errno = EBADF; return EOF; } -#ifdef HAVE_LIBZ - if (fp->compressedfp) + + switch (fp->compression_spec.algorithm) { - result = gzclose(fp->compressedfp); - fp->compressedfp = NULL; - } - else + case PG_COMPRESSION_NONE: + ret = fclose((FILE *) fp->fp); + fp->fp = NULL; + + break; + case PG_COMPRESSION_GZIP: +#ifdef HAVE_LIBZ + ret = gzclose((gzFile) fp->fp); + fp->fp = NULL; +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - { - result = fclose(fp->uncompressedfp); - fp->uncompressedfp = NULL; + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; } + free_keep_errno(fp); - return result; + return ret; } int cfeof(cfp *fp) { + int ret = 0; + + switch (fp->compression_spec.algorithm) + { + case PG_COMPRESSION_NONE: + ret = feof((FILE *) fp->fp); + + break; + case PG_COMPRESSION_GZIP: #ifdef HAVE_LIBZ - if (fp->compressedfp) - return gzeof(fp->compressedfp); - else + ret = gzeof((gzFile) fp->fp); +#else + pg_fatal("this build does not support compression with %s", + "gzip"); #endif - return feof(fp->uncompressedfp); + break; + case PG_COMPRESSION_LZ4: + pg_fatal("compression with %s is not yet supported", "LZ4"); + break; + case PG_COMPRESSION_ZSTD: + pg_fatal("compression with %s is not yet supported", "ZSTD"); + break; + } + + return ret; } const char * get_cfp_error(cfp *fp) { -#ifdef HAVE_LIBZ - if (fp->compressedfp) + if (fp->compression_spec.algorithm == PG_COMPRESSION_GZIP) { +#ifdef HAVE_LIBZ int errnum; - const char *errmsg = gzerror(fp->compressedfp, &errnum); + const char *errmsg = gzerror((gzFile) fp->fp, &errnum); if (errnum != Z_ERRNO) return errmsg; - } +#else + pg_fatal("this build does not support compression with %s", "gzip"); #endif + } + return strerror(errno); } diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index 7f7a0f1ce7..fb94317ad9 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -31,6 +31,7 @@ #endif #include "common/string.h" +#include "compress_io.h" #include "dumputils.h" #include "fe_utils/string_utils.h" #include "lib/stringinfo.h" @@ -43,13 +44,6 @@ #define TEXT_DUMP_HEADER "--\n-- PostgreSQL database dump\n--\n\n" #define TEXT_DUMPALL_HEADER "--\n-- PostgreSQL database cluster dump\n--\n\n" -/* state needed to save/restore an archive's output target */ -typedef struct _outputContext -{ - void *OF; - int gzOut; -} OutputContext; - /* * State for tracking TocEntrys that are ready to process during a parallel * restore. (This used to be a list, and we still call it that, though now @@ -101,8 +95,8 @@ static void dump_lo_buf(ArchiveHandle *AH); static void dumpTimestamp(ArchiveHandle *AH, const char *msg, time_t tim); static void SetOutput(ArchiveHandle *AH, const char *filename, const pg_compress_specification compression_spec); -static OutputContext SaveOutput(ArchiveHandle *AH); -static void RestoreOutput(ArchiveHandle *AH, OutputContext savedContext); +static cfp *SaveOutput(ArchiveHandle *AH); +static void RestoreOutput(ArchiveHandle *AH, cfp *savedOutput); static int restore_toc_entry(ArchiveHandle *AH, TocEntry *te, bool is_parallel); static void restore_toc_entries_prefork(ArchiveHandle *AH, @@ -277,11 +271,8 @@ CloseArchive(Archive *AHX) AH->ClosePtr(AH); /* Close the output */ - errno = 0; /* in case gzclose() doesn't set it */ - if (AH->gzOut) - res = GZCLOSE(AH->OF); - else if (AH->OF != stdout) - res = fclose(AH->OF); + errno = 0; + res = cfclose(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); @@ -362,8 +353,9 @@ RestoreArchive(Archive *AHX) ArchiveHandle *AH = (ArchiveHandle *) AHX; RestoreOptions *ropt = AH->public.ropt; bool parallel_mode; + bool supports_compression; TocEntry *te; - OutputContext sav; + cfp *sav; AH->stage = STAGE_INITIALIZING; @@ -391,17 +383,24 @@ RestoreArchive(Archive *AHX) /* * Make sure we won't need (de)compression we haven't got */ -#ifndef HAVE_LIBZ - if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP && + supports_compression = true; + if (AH->compression_spec.algorithm != PG_COMPRESSION_NONE && + AH->compression_spec.algorithm == PG_COMPRESSION_GZIP && AH->PrintTocDataPtr != NULL) { for (te = AH->toc->next; te != AH->toc; te = te->next) { if (te->hadDumper && (te->reqs & REQ_DATA) != 0) - pg_fatal("cannot restore from compressed archive (compression not supported in this installation)"); + { +#ifndef HAVE_LIBZ + if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) + supports_compression = false; +#endif + if (supports_compression == false) + pg_fatal("cannot restore from compressed archive (compression not supported in this installation)"); + } } } -#endif /* * Prepare index arrays, so we can assume we have them throughout restore. @@ -1128,7 +1127,7 @@ PrintTOCSummary(Archive *AHX) TocEntry *te; pg_compress_specification out_compression_spec = {0}; teSection curSection; - OutputContext sav; + cfp *sav; const char *fmtName; char stamp_str[64]; @@ -1503,58 +1502,32 @@ static void SetOutput(ArchiveHandle *AH, const char *filename, const pg_compress_specification compression_spec) { - int fn; + const char *mode; + int fn = -1; if (filename) { if (strcmp(filename, "-") == 0) fn = fileno(stdout); - else - fn = -1; } else if (AH->FH) fn = fileno(AH->FH); else if (AH->fSpec) { - fn = -1; filename = AH->fSpec; } else fn = fileno(stdout); - /* If compression explicitly requested, use gzopen */ -#ifdef HAVE_LIBZ - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) - { - char fmode[14]; + if (AH->mode == archModeAppend) + mode = PG_BINARY_A; + else + mode = PG_BINARY_W; - /* Don't use PG_BINARY_x since this is zlib */ - sprintf(fmode, "wb%d", compression_spec.level); - if (fn >= 0) - AH->OF = gzdopen(dup(fn), fmode); - else - AH->OF = gzopen(filename, fmode); - AH->gzOut = 1; - } + if (fn >= 0) + AH->OF = cfdopen(dup(fn), mode, compression_spec); else -#endif - { /* Use fopen */ - if (AH->mode == archModeAppend) - { - if (fn >= 0) - AH->OF = fdopen(dup(fn), PG_BINARY_A); - else - AH->OF = fopen(filename, PG_BINARY_A); - } - else - { - if (fn >= 0) - AH->OF = fdopen(dup(fn), PG_BINARY_W); - else - AH->OF = fopen(filename, PG_BINARY_W); - } - AH->gzOut = 0; - } + AH->OF = cfopen(filename, mode, compression_spec); if (!AH->OF) { @@ -1565,33 +1538,24 @@ SetOutput(ArchiveHandle *AH, const char *filename, } } -static OutputContext +static cfp * SaveOutput(ArchiveHandle *AH) { - OutputContext sav; - - sav.OF = AH->OF; - sav.gzOut = AH->gzOut; - - return sav; + return (cfp *) AH->OF; } static void -RestoreOutput(ArchiveHandle *AH, OutputContext savedContext) +RestoreOutput(ArchiveHandle *AH, cfp *savedOutput) { int res; - errno = 0; /* in case gzclose() doesn't set it */ - if (AH->gzOut) - res = GZCLOSE(AH->OF); - else - res = fclose(AH->OF); + errno = 0; + res = cfclose(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); - AH->gzOut = savedContext.gzOut; - AH->OF = savedContext.OF; + AH->OF = savedOutput; } @@ -1715,22 +1679,17 @@ ahwrite(const void *ptr, size_t size, size_t nmemb, ArchiveHandle *AH) bytes_written = size * nmemb; } - else if (AH->gzOut) - bytes_written = GZWRITE(ptr, size, nmemb, AH->OF); else if (AH->CustomOutPtr) bytes_written = AH->CustomOutPtr(AH, ptr, size * nmemb); + /* + * If we're doing a restore, and it's direct to DB, and we're connected + * then send it to the DB. + */ + else if (RestoringToDB(AH)) + bytes_written = ExecuteSqlCommandBuf(&AH->public, (const char *) ptr, size * nmemb); else - { - /* - * If we're doing a restore, and it's direct to DB, and we're - * connected then send it to the DB. - */ - if (RestoringToDB(AH)) - bytes_written = ExecuteSqlCommandBuf(&AH->public, (const char *) ptr, size * nmemb); - else - bytes_written = fwrite(ptr, size, nmemb, AH->OF) * size; - } + bytes_written = cfwrite(ptr, size * nmemb, AH->OF); if (bytes_written != size * nmemb) WRITE_ERROR_EXIT; @@ -2219,6 +2178,7 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, SetupWorkerPtrType setupWorkerPtr) { ArchiveHandle *AH; + pg_compress_specification out_compress_spec = {0}; pg_log_debug("allocating AH for %s, format %d", FileSpec ? FileSpec : "(stdio)", fmt); @@ -2272,8 +2232,8 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, memset(&(AH->sqlparse), 0, sizeof(AH->sqlparse)); /* Open stdout with no compression for AH output handle */ - AH->gzOut = 0; - AH->OF = stdout; + out_compress_spec.algorithm = PG_COMPRESSION_NONE; + AH->OF = cfdopen(dup(fileno(stdout)), PG_BINARY_A, out_compress_spec); /* * On Windows, we need to use binary mode to read/write non-text files, diff --git a/src/bin/pg_dump/pg_backup_archiver.h b/src/bin/pg_dump/pg_backup_archiver.h index f72446ed5b..4725e49747 100644 --- a/src/bin/pg_dump/pg_backup_archiver.h +++ b/src/bin/pg_dump/pg_backup_archiver.h @@ -32,30 +32,6 @@ #define LOBBUFSIZE 16384 -#ifdef HAVE_LIBZ -#include <zlib.h> -#define GZCLOSE(fh) gzclose(fh) -#define GZWRITE(p, s, n, fh) gzwrite(fh, p, (n) * (s)) -#define GZREAD(p, s, n, fh) gzread(fh, p, (n) * (s)) -#define GZEOF(fh) gzeof(fh) -#else -#define GZCLOSE(fh) fclose(fh) -#define GZWRITE(p, s, n, fh) (fwrite(p, s, n, fh) * (s)) -#define GZREAD(p, s, n, fh) fread(p, s, n, fh) -#define GZEOF(fh) feof(fh) -/* this is just the redefinition of a libz constant */ -#define Z_DEFAULT_COMPRESSION (-1) - -typedef struct _z_stream -{ - void *next_in; - void *next_out; - size_t avail_in; - size_t avail_out; -} z_stream; -typedef z_stream *z_streamp; -#endif - /* Data block types */ #define BLK_DATA 1 #define BLK_BLOBS 3 @@ -319,8 +295,7 @@ struct _archiveHandle char *fSpec; /* Archive File Spec */ FILE *FH; /* General purpose file handle */ - void *OF; - int gzOut; /* Output file */ + void *OF; /* Output file */ struct _tocEntry *toc; /* Header of circular list of TOC entries */ int tocCount; /* Number of TOC entries */ -- 2.34.1 [text/x-patch] v20-0003-THIS-IS-WIP-PATCH-PRESENTED-HERE-FOR-COMPLETENES.patch (29.0K, ../../twKgLFeqHEshCjdrbnzkwTMHiqfqvrrPSni7eKZjcxkForO-sI7HL4AKFgk9En1nGk3i4TXOY_An2xnCzv8MNjoLz4iupACaRB-0VRV20V4=@pm.me/3-v20-0003-THIS-IS-WIP-PATCH-PRESENTED-HERE-FOR-COMPLETENES.patch) download | inline diff: From 3c7e5b8ebe3c4d063f8d281a4c9d854fef6578b0 Mon Sep 17 00:00:00 2001 From: Georgios Kokolatos <[email protected]> Date: Mon, 16 Jan 2023 15:01:05 +0000 Subject: [PATCH v20 3/3] THIS IS WIP PATCH PRESENTED HERE FOR COMPLETENESS: LZ4 compression Please be aware that comments, references, naming, etc, are not scrutinized. Those will be addressed once 0001 of the series is getting closer to a commitable state. However the patch should apply cleanly and the tests should be passing. Within compress_lz4.{c,h} the streaming API and a file API compression is implemented.. The first one, is aimed at inlined use cases and thus simple lz4.h calls can be used directly. The second one is generating output, or is parsing input, which can be read/generated via the lz4 utility. Wherever the LZ4F api does not implement all the functionality corresponding to fread(), fwrite(), fgets(), fgetc(), feof(), and fclose(), it has been implemented localy. --- doc/src/sgml/ref/pg_dump.sgml | 13 +- src/bin/pg_dump/Makefile | 2 + src/bin/pg_dump/compress_io.c | 11 +- src/bin/pg_dump/compress_lz4.c | 618 +++++++++++++++++++++++++++ src/bin/pg_dump/compress_lz4.h | 22 + src/bin/pg_dump/meson.build | 8 +- src/bin/pg_dump/pg_backup_archiver.c | 14 +- src/bin/pg_dump/pg_dump.c | 5 +- src/bin/pg_dump/t/002_pg_dump.pl | 82 +++- src/tools/pginclude/cpluspluscheck | 1 + src/tools/pgindent/typedefs.list | 1 + 11 files changed, 756 insertions(+), 21 deletions(-) create mode 100644 src/bin/pg_dump/compress_lz4.c create mode 100644 src/bin/pg_dump/compress_lz4.h diff --git a/doc/src/sgml/ref/pg_dump.sgml b/doc/src/sgml/ref/pg_dump.sgml index 2c938cd7e1..49d218905f 100644 --- a/doc/src/sgml/ref/pg_dump.sgml +++ b/doc/src/sgml/ref/pg_dump.sgml @@ -330,9 +330,10 @@ PostgreSQL documentation machine-readable format that <application>pg_restore</application> can read. A directory format archive can be manipulated with standard Unix tools; for example, files in an uncompressed archive - can be compressed with the <application>gzip</application> tool. - This format is compressed by default and also supports parallel - dumps. + can be compressed with the <application>gzip</application> or + <application>lz4</application>tool. + This format is compressed by default using <literal>gzip</literal> + and also supports parallel dumps. </para> </listitem> </varlistentry> @@ -654,7 +655,7 @@ PostgreSQL documentation <para> Specify the compression method and/or the compression level to use. The compression method can be set to <literal>gzip</literal> or - <literal>none</literal> for no compression. + <literal>lz4</literal> or <literal>none</literal> for no compression. A compression detail string can optionally be specified. If the detail string is an integer, it specifies the compression level. Otherwise, it should be a comma-separated list of items, each of the @@ -675,8 +676,8 @@ PostgreSQL documentation individual table-data segments, and the default is to compress using <literal>gzip</literal> at a moderate level. For plain text output, setting a nonzero compression level causes the entire output file to be compressed, - as though it had been fed through <application>gzip</application>; but the default - is not to compress. + as though it had been fed through <application>gzip</application> or + <application>lz4</application>; but the default is not to compress. </para> <para> The tar archive format currently does not support compression at all. diff --git a/src/bin/pg_dump/Makefile b/src/bin/pg_dump/Makefile index 7a19f5d617..a1401377ab 100644 --- a/src/bin/pg_dump/Makefile +++ b/src/bin/pg_dump/Makefile @@ -17,6 +17,7 @@ top_builddir = ../../.. include $(top_builddir)/src/Makefile.global export GZIP_PROGRAM=$(GZIP) +export LZ4 export with_icu override CPPFLAGS := -I$(libpq_srcdir) $(CPPFLAGS) @@ -25,6 +26,7 @@ LDFLAGS_INTERNAL += -L$(top_builddir)/src/fe_utils -lpgfeutils $(libpq_pgport) OBJS = \ $(WIN32RES) \ compress_gzip.o \ + compress_lz4.o \ compress_io.o \ dumputils.o \ parallel.o \ diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index 7274039e87..43bd4c1a2c 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -60,6 +60,7 @@ #include "compress_io.h" #include "compress_gzip.h" +#include "compress_lz4.h" #include "pg_backup_utils.h" #ifdef HAVE_LIBZ @@ -136,7 +137,7 @@ AllocateCompressor(const pg_compress_specification compression_spec, InitCompressorGzip(cs, compression_spec); break; case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); + InitCompressorLZ4(cs, compression_spec); break; case PG_COMPRESSION_ZSTD: pg_fatal("compression with %s is not yet supported", "ZSTD"); @@ -326,7 +327,7 @@ InitCompressFileHandle(const pg_compress_specification compression_spec) InitCompressGzip(CFH, compression_spec); break; case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); + InitCompressLZ4(CFH, compression_spec); break; case PG_COMPRESSION_ZSTD: pg_fatal("compression with %s is not yet supported", "ZSTD"); @@ -340,12 +341,12 @@ InitCompressFileHandle(const pg_compress_specification compression_spec) * Open a file for reading. 'path' is the file to open, and 'mode' should * be either "r" or "rb". * - * If the file at 'path' does not exist, we append the "{.gz,.lz4}" suffix (i + * If the file at 'path' does not exist, we append the ".gz" suffix (if * 'path' doesn't already have it) and try again. So if you pass "foo" as - * 'path', this will open either "foo" or "foo.gz" or "foo.lz4", trying in that - * order. + * 'path', this will open either "foo" or "foo.gz", trying in that order. * * On failure, return NULL with an error code in errno. + * */ CompressFileHandle * InitDiscoverCompressFileHandle(const char *path, const char *mode) diff --git a/src/bin/pg_dump/compress_lz4.c b/src/bin/pg_dump/compress_lz4.c new file mode 100644 index 0000000000..c97e16187a --- /dev/null +++ b/src/bin/pg_dump/compress_lz4.c @@ -0,0 +1,618 @@ +/*------------------------------------------------------------------------- + * + * compress_lz4.c + * Routines for archivers to write an uncompressed or compressed data + * stream. + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_lz4.c + * + *------------------------------------------------------------------------- + */ +#include "postgres_fe.h" +#include "pg_backup_utils.h" + +#include "compress_lz4.h" + +#ifdef USE_LZ4 +#include <lz4.h> +#include <lz4frame.h> + +#define LZ4_OUT_SIZE (4 * 1024) +#define LZ4_IN_SIZE (16 * 1024) + +/* + * LZ4F_HEADER_SIZE_MAX first appeared in v1.7.5 of the library. + * Redefine it for installations with a lesser version. + */ +#ifndef LZ4F_HEADER_SIZE_MAX +#define LZ4F_HEADER_SIZE_MAX 32 +#endif + +/*---------------------- + * Compressor API + *---------------------- + */ + +typedef struct LZ4CompressorState +{ + char *outbuf; + size_t outsize; +} LZ4CompressorState; + +/* Private routines that support LZ4 compressed data I/O */ +static void ReadDataFromArchiveLZ4(ArchiveHandle *AH, CompressorState *cs); +static void WriteDataToArchiveLZ4(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen); +static void EndCompressorLZ4(ArchiveHandle *AH, CompressorState *cs); + +static void +ReadDataFromArchiveLZ4(ArchiveHandle *AH, CompressorState *cs) +{ + LZ4_streamDecode_t lz4StreamDecode; + char *buf; + char *decbuf; + size_t buflen; + size_t cnt; + + buflen = LZ4_IN_SIZE; + buf = pg_malloc(buflen); + decbuf = pg_malloc(buflen); + + LZ4_setStreamDecode(&lz4StreamDecode, NULL, 0); + + while ((cnt = cs->readF(AH, &buf, &buflen))) + { + int decBytes = LZ4_decompress_safe_continue(&lz4StreamDecode, + buf, decbuf, + cnt, buflen); + + ahwrite(decbuf, 1, decBytes, AH); + } + + pg_free(buf); + pg_free(decbuf); +} + +static void +WriteDataToArchiveLZ4(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen) +{ + LZ4CompressorState *LZ4cs = (LZ4CompressorState *) cs->private_data; + size_t compressed; + size_t requiredsize = LZ4_compressBound(dLen); + + if (requiredsize > LZ4cs->outsize) + { + LZ4cs->outbuf = pg_realloc(LZ4cs->outbuf, requiredsize); + LZ4cs->outsize = requiredsize; + } + + compressed = LZ4_compress_default(data, LZ4cs->outbuf, + dLen, LZ4cs->outsize); + + if (compressed <= 0) + pg_fatal("failed to LZ4 compress data"); + + cs->writeF(AH, LZ4cs->outbuf, compressed); +} + +static void +EndCompressorLZ4(ArchiveHandle *AH, CompressorState *cs) +{ + LZ4CompressorState *LZ4cs; + + LZ4cs = (LZ4CompressorState *) cs->private_data; + if (LZ4cs) + { + pg_free(LZ4cs->outbuf); + pg_free(LZ4cs); + cs->private_data = NULL; + } +} + + +/* Public routines that support LZ4 compressed data I/O */ +void +InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec) +{ + cs->readData = ReadDataFromArchiveLZ4; + cs->writeData = WriteDataToArchiveLZ4; + cs->end = EndCompressorLZ4; + + cs->compression_spec = compression_spec; + + /* Will be lazy init'd */ + cs->private_data = pg_malloc0(sizeof(LZ4CompressorState)); +} + +/*---------------------- + * Compress File API + *---------------------- + */ + +/* + * State needed for LZ4 (de)compression using the CompressFileHandle API. + */ +typedef struct LZ4File +{ + FILE *fp; + + LZ4F_preferences_t prefs; + + LZ4F_compressionContext_t ctx; + LZ4F_decompressionContext_t dtx; + + bool inited; + bool compressing; + + size_t buflen; + char *buffer; + + size_t overflowalloclen; + size_t overflowlen; + char *overflowbuf; + + size_t errcode; +} LZ4File; + +/* + * LZ4 equivalent to feof() or gzeof(). The end of file + * is reached if there is no decompressed output in the + * overflow buffer and the end of the file is reached. + */ +static int +LZ4File_eof(CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + + return fs->overflowlen == 0 && feof(fs->fp); +} + +static const char * +LZ4File_get_error(CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + const char *errmsg; + + if (LZ4F_isError(fs->errcode)) + errmsg = LZ4F_getErrorName(fs->errcode); + else + errmsg = strerror(errno); + + return errmsg; +} + +/* + * Prepare an already alloc'ed LZ4File struct for subsequent calls. + * + * It creates the nessary contexts for the operations. When compressing, + * it additionally writes the LZ4 header in the output stream. + */ +static int +LZ4File_init(LZ4File * fs, int size, bool compressing) +{ + size_t status; + + if (fs->inited) + return 0; + + fs->compressing = compressing; + fs->inited = true; + + if (fs->compressing) + { + fs->buflen = LZ4F_compressBound(LZ4_IN_SIZE, &fs->prefs); + if (fs->buflen < LZ4F_HEADER_SIZE_MAX) + fs->buflen = LZ4F_HEADER_SIZE_MAX; + + status = LZ4F_createCompressionContext(&fs->ctx, LZ4F_VERSION); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return 1; + } + + fs->buffer = pg_malloc(fs->buflen); + status = LZ4F_compressBegin(fs->ctx, fs->buffer, fs->buflen, + &fs->prefs); + + if (LZ4F_isError(status)) + { + fs->errcode = status; + return 1; + } + + if (fwrite(fs->buffer, 1, status, fs->fp) != status) + { + errno = errno ? : ENOSPC; + return 1; + } + } + else + { + status = LZ4F_createDecompressionContext(&fs->dtx, LZ4F_VERSION); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return 1; + } + + fs->buflen = size > LZ4_OUT_SIZE ? size : LZ4_OUT_SIZE; + fs->buffer = pg_malloc(fs->buflen); + + fs->overflowalloclen = fs->buflen; + fs->overflowbuf = pg_malloc(fs->overflowalloclen); + fs->overflowlen = 0; + } + + return 0; +} + +/* + * Read already decompressed content from the overflow buffer into 'ptr' up to + * 'size' bytes, if available. If the eol_flag is set, then stop at the first + * occurance of the new line char prior to 'size' bytes. + * + * Any unread content in the overflow buffer, is moved to the beginning. + */ +static int +LZ4File_read_overflow(LZ4File * fs, void *ptr, int size, bool eol_flag) +{ + char *p; + int readlen = 0; + + if (fs->overflowlen == 0) + return 0; + + if (fs->overflowlen >= size) + readlen = size; + else + readlen = fs->overflowlen; + + if (eol_flag && (p = memchr(fs->overflowbuf, '\n', readlen))) + /* Include the line terminating char */ + readlen = p - fs->overflowbuf + 1; + + memcpy(ptr, fs->overflowbuf, readlen); + fs->overflowlen -= readlen; + + if (fs->overflowlen > 0) + memmove(fs->overflowbuf, fs->overflowbuf + readlen, fs->overflowlen); + + return readlen; +} + +/* + * The workhorse for reading decompressed content out of an LZ4 compressed + * stream. + * + * It will read up to 'ptrsize' decompressed content, or up to the new line char + * if found first when the eol_flag is set. It is possible that the decompressed + * output generated by reading any compressed input via the LZ4F API, exceeds + * 'ptrsize'. Any exceeding decompressed content is stored at an overflow + * buffer within LZ4File. Of course, when the function is called, it will first + * try to consume any decompressed content already present in the overflow + * buffer, before decompressing new content. + */ +static int +LZ4File_read_internal(LZ4File * fs, void *ptr, int ptrsize, bool eol_flag) +{ + size_t dsize = 0; + size_t rsize; + size_t size = ptrsize; + bool eol_found = false; + + void *readbuf; + + /* Lazy init */ + if (!fs->inited && LZ4File_init(fs, size, false /* decompressing */ )) + return -1; + + /* Verfiy that there is enough space in the outbuf */ + if (size > fs->buflen) + { + fs->buflen = size; + fs->buffer = pg_realloc(fs->buffer, size); + } + + /* use already decompressed content if available */ + dsize = LZ4File_read_overflow(fs, ptr, size, eol_flag); + if (dsize == size || (eol_flag && memchr(ptr, '\n', dsize))) + return dsize; + + readbuf = pg_malloc(size); + + do + { + char *rp; + char *rend; + + rsize = fread(readbuf, 1, size, fs->fp); + if (rsize < size && !feof(fs->fp)) + return -1; + + rp = (char *) readbuf; + rend = (char *) readbuf + rsize; + + while (rp < rend) + { + size_t status; + size_t outlen = fs->buflen; + size_t read_remain = rend - rp; + + memset(fs->buffer, 0, outlen); + status = LZ4F_decompress(fs->dtx, fs->buffer, &outlen, + rp, &read_remain, NULL); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return -1; + } + + rp += read_remain; + + /* + * fill in what space is available in ptr if the eol flag is set, + * either skip if one already found or fill up to EOL if present + * in the outbuf + */ + if (outlen > 0 && dsize < size && eol_found == false) + { + char *p; + size_t lib = (eol_flag == 0) ? size - dsize : size - 1 - dsize; + size_t len = outlen < lib ? outlen : lib; + + if (eol_flag == true && + (p = memchr(fs->buffer, '\n', outlen)) && + (size_t) (p - fs->buffer + 1) <= len) + { + len = p - fs->buffer + 1; + eol_found = true; + } + + memcpy((char *) ptr + dsize, fs->buffer, len); + dsize += len; + + /* move what did not fit, if any, at the begining of the buf */ + if (len < outlen) + memmove(fs->buffer, fs->buffer + len, outlen - len); + outlen -= len; + } + + /* if there is available output, save it */ + if (outlen > 0) + { + while (fs->overflowlen + outlen > fs->overflowalloclen) + { + fs->overflowalloclen *= 2; + fs->overflowbuf = pg_realloc(fs->overflowbuf, + fs->overflowalloclen); + } + + memcpy(fs->overflowbuf + fs->overflowlen, fs->buffer, outlen); + fs->overflowlen += outlen; + } + } + } while (rsize == size && dsize < size && eol_found == 0); + + pg_free(readbuf); + + return (int) dsize; +} + +/* + * Compress size bytes from ptr and write them to the stream. + */ +static size_t +LZ4File_write(const void *ptr, size_t size, CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + size_t status; + int remaining = size; + + if (!fs->inited && LZ4File_init(fs, size, true)) + return -1; + + while (remaining > 0) + { + int chunk = remaining < LZ4_IN_SIZE ? remaining : LZ4_IN_SIZE; + + remaining -= chunk; + + status = LZ4F_compressUpdate(fs->ctx, fs->buffer, fs->buflen, + ptr, chunk, NULL); + if (LZ4F_isError(status)) + { + fs->errcode = status; + return -1; + } + + if (fwrite(fs->buffer, 1, status, fs->fp) != status) + { + errno = errno ? : ENOSPC; + return 1; + } + } + + return size; +} + +/* + * fread() equivalent implementation for LZ4 compressed files. + */ +static size_t +LZ4File_read(void *ptr, size_t size, CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + int ret; + + ret = LZ4File_read_internal(fs, ptr, size, false); + if (ret != size && !LZ4File_eof(CFH)) + pg_fatal("could not read from input file: %s", LZ4File_get_error(CFH)); + + return ret; +} + +/* + * fgetc() equivalent implementation for LZ4 compressed files. + */ +static int +LZ4File_getc(CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + unsigned char c; + + if (LZ4File_read_internal(fs, &c, 1, false) != 1) + { + if (!LZ4File_eof(CFH)) + pg_fatal("could not read from input file: %s", LZ4File_get_error(CFH)); + else + pg_fatal("could not read from input file: end of file"); + } + + return c; +} + +/* + * fgets() equivalent implementation for LZ4 compressed files. + */ +static char * +LZ4File_gets(char *ptr, int size, CompressFileHandle *CFH) +{ + LZ4File *fs = (LZ4File *) CFH->private_data; + size_t dsize; + + dsize = LZ4File_read_internal(fs, ptr, size, true); + if (dsize < 0) + pg_fatal("could not read from input file: %s", LZ4File_get_error(CFH)); + + /* Done reading */ + if (dsize == 0) + return NULL; + + return ptr; +} + +/* + * Finalize (de)compression of a stream. When compressing it will write any + * remaining content and/or generated footer from the LZ4 API. + */ +static int +LZ4File_close(CompressFileHandle *CFH) +{ + FILE *fp; + LZ4File *fs = (LZ4File *) CFH->private_data; + size_t status; + int ret; + + fp = fs->fp; + if (fs->inited) + { + if (fs->compressing) + { + status = LZ4F_compressEnd(fs->ctx, fs->buffer, fs->buflen, NULL); + if (LZ4F_isError(status)) + pg_fatal("failed to end compression: %s", + LZ4F_getErrorName(status)); + else if ((ret = fwrite(fs->buffer, 1, status, fs->fp)) != status) + { + errno = errno ? : ENOSPC; + WRITE_ERROR_EXIT; + } + + status = LZ4F_freeCompressionContext(fs->ctx); + if (LZ4F_isError(status)) + pg_fatal("failed to end compression: %s", + LZ4F_getErrorName(status)); + } + else + { + status = LZ4F_freeDecompressionContext(fs->dtx); + if (LZ4F_isError(status)) + pg_fatal("failed to end decompression: %s", + LZ4F_getErrorName(status)); + pg_free(fs->overflowbuf); + } + + pg_free(fs->buffer); + } + + pg_free(fs); + + return fclose(fp); +} + +static int +LZ4File_open(const char *path, int fd, const char *mode, + CompressFileHandle *CFH) +{ + FILE *fp; + LZ4File *lz4fp = (LZ4File *) CFH->private_data; + + if (fd >= 0) + fp = fdopen(fd, mode); + else + fp = fopen(path, mode); + if (fp == NULL) + { + lz4fp->errcode = errno; + return 1; + } + + lz4fp->fp = fp; + + return 0; +} + +static int +LZ4File_open_write(const char *path, const char *mode, CompressFileHandle *CFH) +{ + char *fname; + int ret; + + fname = psprintf("%s.lz4", path); + ret = CFH->open_func(fname, -1, mode, CFH); + pg_free(fname); + + return ret; +} + +void +InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +{ + LZ4File *lz4fp; + + CFH->open_func = LZ4File_open; + CFH->open_write_func = LZ4File_open_write; + CFH->read_func = LZ4File_read; + CFH->write_func = LZ4File_write; + CFH->gets_func = LZ4File_gets; + CFH->getc_func = LZ4File_getc; + CFH->eof_func = LZ4File_eof; + CFH->close_func = LZ4File_close; + CFH->get_error_func = LZ4File_get_error; + + CFH->compression_spec = compression_spec; + lz4fp = pg_malloc0(sizeof(*lz4fp)); + if (CFH->compression_spec.level >= 0) + lz4fp->prefs.compressionLevel = CFH->compression_spec.level; + + CFH->private_data = lz4fp; +} +#else /* USE_LZ4 */ +void +InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec) +{ + pg_fatal("this build does not support compression with %s", "LZ4"); +} + +void +InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +{ + pg_fatal("this build does not support compression with %s", "LZ4"); +} +#endif /* USE_LZ4 */ diff --git a/src/bin/pg_dump/compress_lz4.h b/src/bin/pg_dump/compress_lz4.h new file mode 100644 index 0000000000..74595db1b9 --- /dev/null +++ b/src/bin/pg_dump/compress_lz4.h @@ -0,0 +1,22 @@ +/*------------------------------------------------------------------------- + * + * compress_lz4.h + * Interface to compress_io.c routines + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_lz4.h + * + *------------------------------------------------------------------------- + */ +#ifndef _COMPRESS_LZ4_H_ +#define _COMPRESS_LZ4_H_ + +#include "compress_io.h" + +extern void InitCompressorLZ4(CompressorState *cs, const pg_compress_specification compression_spec); +extern void InitCompressLZ4(CompressFileHandle *CFH, const pg_compress_specification compression_spec); + +#endif /* _COMPRESS_LZ4_H_ */ diff --git a/src/bin/pg_dump/meson.build b/src/bin/pg_dump/meson.build index aa2c91829c..473d40d456 100644 --- a/src/bin/pg_dump/meson.build +++ b/src/bin/pg_dump/meson.build @@ -3,6 +3,7 @@ pg_dump_common_sources = files( 'compress_io.c', 'compress_gzip.c', + 'compress_lz4.c', 'dumputils.c', 'parallel.c', 'pg_backup_archiver.c', @@ -17,7 +18,7 @@ pg_dump_common_sources = files( pg_dump_common = static_library('libpgdump_common', pg_dump_common_sources, c_pch: pch_postgres_fe_h, - dependencies: [frontend_code, libpq, zlib], + dependencies: [frontend_code, libpq, lz4, zlib], kwargs: internal_lib_args, ) @@ -85,7 +86,10 @@ tests += { 'sd': meson.current_source_dir(), 'bd': meson.current_build_dir(), 'tap': { - 'env': {'GZIP_PROGRAM': gzip.path()}, + 'env': { + 'GZIP_PROGRAM': gzip.path(), + 'LZ4': program_lz4.found() ? program_lz4.path() : '', + }, 'tests': [ 't/001_basic.pl', 't/002_pg_dump.pl', diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index 1f207c6f4d..119b7f2553 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -395,6 +395,10 @@ RestoreArchive(Archive *AHX) #ifndef HAVE_LIBZ if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) supports_compression = false; +#endif +#ifndef USE_LZ4 + if (AH->compression_spec.algorithm == PG_COMPRESSION_LZ4) + supports_compression = false; #endif if (supports_compression == false) pg_fatal("cannot restore from compressed archive (compression not supported in this installation)"); @@ -2074,7 +2078,7 @@ _discoverArchiveFormat(ArchiveHandle *AH) /* * Check if the specified archive is a directory. If so, check if - * there's a "toc.dat" (or "toc.dat.gz") file in it. + * there's a "toc.dat" (or "toc.dat.{gz,lz4}") file in it. */ if (stat(AH->fSpec, &st) == 0 && S_ISDIR(st.st_mode)) { @@ -2084,6 +2088,10 @@ _discoverArchiveFormat(ArchiveHandle *AH) #ifdef HAVE_LIBZ if (_fileExistsInDirectory(AH->fSpec, "toc.dat.gz")) return AH->format; +#endif +#ifdef USE_LZ4 + if (_fileExistsInDirectory(AH->fSpec, "toc.dat.lz4")) + return AH->format; #endif pg_fatal("directory \"%s\" does not appear to be a valid archive (\"toc.dat\" does not exist)", AH->fSpec); @@ -3747,6 +3755,10 @@ ReadHead(ArchiveHandle *AH) #ifndef HAVE_LIBZ if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) unsupported = true; +#endif +#ifndef USE_LZ4 + if (AH->compression_spec.algorithm == PG_COMPRESSION_LZ4) + unsupported = true; #endif if (unsupported) pg_fatal("archive is compressed, but this installation does not support compression"); diff --git a/src/bin/pg_dump/pg_dump.c b/src/bin/pg_dump/pg_dump.c index 2c0a969972..f392760e06 100644 --- a/src/bin/pg_dump/pg_dump.c +++ b/src/bin/pg_dump/pg_dump.c @@ -715,13 +715,12 @@ main(int argc, char **argv) case PG_COMPRESSION_NONE: /* fallthrough */ case PG_COMPRESSION_GZIP: + /* fallthrough */ + case PG_COMPRESSION_LZ4: break; case PG_COMPRESSION_ZSTD: pg_fatal("compression with %s is not yet supported", "ZSTD"); break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; } /* diff --git a/src/bin/pg_dump/t/002_pg_dump.pl b/src/bin/pg_dump/t/002_pg_dump.pl index f3ba926321..f497ec6040 100644 --- a/src/bin/pg_dump/t/002_pg_dump.pl +++ b/src/bin/pg_dump/t/002_pg_dump.pl @@ -139,6 +139,80 @@ my %pgdump_runs = ( args => [ '-d', "$tempdir/compression_gzip_plain.sql.gz", ], }, }, + + # Do not use --no-sync to give test coverage for data sync. + compression_lz4_custom => { + test_key => 'compression', + compile_option => 'lz4', + dump_cmd => [ + 'pg_dump', '--format=custom', + '--compress=lz4', "--file=$tempdir/compression_lz4_custom.dump", + 'postgres', + ], + restore_cmd => [ + 'pg_restore', + "--file=$tempdir/compression_lz4_custom.sql", + "$tempdir/compression_lz4_custom.dump", + ], + command_like => { + command => [ + 'pg_restore', + '-l', "$tempdir/compression_lz4_custom.dump", + ], + expected => qr/Compression: lz4/, + name => 'data content is lz4 compressed' + }, + }, + + # Do not use --no-sync to give test coverage for data sync. + compression_lz4_dir => { + test_key => 'compression', + compile_option => 'lz4', + dump_cmd => [ + 'pg_dump', '--jobs=2', + '--format=directory', '--compress=lz4:1', + "--file=$tempdir/compression_lz4_dir", 'postgres', + ], + # Give coverage for manually compressed blob.toc files during + # restore. + compress_cmd => { + program => $ENV{'LZ4'}, + args => [ + '-z', '-f', '--rm', + "$tempdir/compression_lz4_dir/blobs.toc", + "$tempdir/compression_lz4_dir/blobs.toc.lz4", + ], + }, + # Verify that data files where compressed + glob_patterns => [ + "$tempdir/compression_lz4_dir/toc.dat", + "$tempdir/compression_lz4_dir/*.dat.lz4", + ], + restore_cmd => [ + 'pg_restore', '--jobs=2', + "--file=$tempdir/compression_lz4_dir.sql", + "$tempdir/compression_lz4_dir", + ], + }, + + compression_lz4_plain => { + test_key => 'compression', + compile_option => 'lz4', + dump_cmd => [ + 'pg_dump', '--format=plain', '--compress=lz4', + "--file=$tempdir/compression_lz4_plain.sql.lz4", 'postgres', + ], + # Decompress the generated file to run through the tests. + compress_cmd => { + program => $ENV{'LZ4'}, + args => [ + '-d', '-f', + "$tempdir/compression_lz4_plain.sql.lz4", + "$tempdir/compression_lz4_plain.sql", + ], + }, + }, + clean => { dump_cmd => [ 'pg_dump', @@ -4175,11 +4249,11 @@ foreach my $run (sort keys %pgdump_runs) my $run_db = 'postgres'; # Skip command-level tests for gzip if there is no support for it. - if ( defined($pgdump_runs{$run}->{compile_option}) - && $pgdump_runs{$run}->{compile_option} eq 'gzip' - && !$supports_gzip) + if ($pgdump_runs{$run}->{compile_option} && + ($pgdump_runs{$run}->{compile_option} eq 'gzip' && !$supports_gzip) || + ($pgdump_runs{$run}->{compile_option} eq 'lz4' && !$supports_lz4)) { - note "$run: skipped due to no gzip support"; + note "$run: skipped due to no $pgdump_runs{$run}->{compile_option} support"; next; } diff --git a/src/tools/pginclude/cpluspluscheck b/src/tools/pginclude/cpluspluscheck index da8e4861f4..1f7f817e4d 100755 --- a/src/tools/pginclude/cpluspluscheck +++ b/src/tools/pginclude/cpluspluscheck @@ -152,6 +152,7 @@ do # as field names, which is unfortunate but we won't change it now. test "$f" = src/bin/pg_dump/compress_gzip.h && continue test "$f" = src/bin/pg_dump/compress_io.h && continue + test "$f" = src/bin/pg_dump/compress_lz4.h && continue test "$f" = src/bin/pg_dump/parallel.h && continue test "$f" = src/bin/pg_dump/pg_backup_archiver.h && continue test "$f" = src/bin/pg_dump/pg_dump.h && continue diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 840191d680..232228d427 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -1381,6 +1381,7 @@ LWLock LWLockHandle LWLockMode LWLockPadded +LZ4CompressorState LZ4F_compressionContext_t LZ4F_decompressOptions_t LZ4F_decompressionContext_t -- 2.34.1 [text/x-patch] v20-0002-THIS-IS-WIP-PATCH-PRESENTED-HERE-FOR-COMPLETENES.patch (65.5K, ../../twKgLFeqHEshCjdrbnzkwTMHiqfqvrrPSni7eKZjcxkForO-sI7HL4AKFgk9En1nGk3i4TXOY_An2xnCzv8MNjoLz4iupACaRB-0VRV20V4=@pm.me/4-v20-0002-THIS-IS-WIP-PATCH-PRESENTED-HERE-FOR-COMPLETENES.patch) download | inline diff: From 7c27acd1faaf4a640a1d0ba7989dec39c853660f Mon Sep 17 00:00:00 2001 From: Georgios Kokolatos <[email protected]> Date: Mon, 16 Jan 2023 14:59:41 +0000 Subject: [PATCH v20 2/3] THIS IS WIP PATCH PRESENTED HERE FOR COMPLETENESS: Compressor API Please be aware that comments, references, naming, etc, are not scrutinized. Those will be addressed once 0001 of the series is getting closer to a commitable state. However the patch should apply cleanly and the tests should be passing. The purpose of this API is to allow for easier addition of new compression methods. CompressFileHandle is substituting the cfp* family of functions under a struct of function pointers for opening, writing, etc. The implementor of a new compression method is now able to "simply" just add those definitions. Custom compressed archives now need to store the compression algorithm in their header. This requires a bump in the version number. The level of compression is no longer stored in the dump as it is irrelevant. --- src/bin/pg_dump/Makefile | 1 + src/bin/pg_dump/compress_gzip.c | 405 ++++++++++++ src/bin/pg_dump/compress_gzip.h | 22 + src/bin/pg_dump/compress_io.c | 914 +++++++------------------- src/bin/pg_dump/compress_io.h | 68 +- src/bin/pg_dump/meson.build | 1 + src/bin/pg_dump/pg_backup_archiver.c | 102 +-- src/bin/pg_dump/pg_backup_archiver.h | 5 +- src/bin/pg_dump/pg_backup_custom.c | 23 +- src/bin/pg_dump/pg_backup_directory.c | 94 +-- src/bin/pg_dump/t/002_pg_dump.pl | 10 +- src/tools/pginclude/cpluspluscheck | 1 + src/tools/pgindent/typedefs.list | 2 + 13 files changed, 851 insertions(+), 797 deletions(-) create mode 100644 src/bin/pg_dump/compress_gzip.c create mode 100644 src/bin/pg_dump/compress_gzip.h diff --git a/src/bin/pg_dump/Makefile b/src/bin/pg_dump/Makefile index ef1ed0f3e5..7a19f5d617 100644 --- a/src/bin/pg_dump/Makefile +++ b/src/bin/pg_dump/Makefile @@ -24,6 +24,7 @@ LDFLAGS_INTERNAL += -L$(top_builddir)/src/fe_utils -lpgfeutils $(libpq_pgport) OBJS = \ $(WIN32RES) \ + compress_gzip.o \ compress_io.o \ dumputils.o \ parallel.o \ diff --git a/src/bin/pg_dump/compress_gzip.c b/src/bin/pg_dump/compress_gzip.c new file mode 100644 index 0000000000..60fb95d7b7 --- /dev/null +++ b/src/bin/pg_dump/compress_gzip.c @@ -0,0 +1,405 @@ +/*------------------------------------------------------------------------- + * + * compress_gzip.c + * Routines for archivers to write an uncompressed or compressed data + * stream. + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_gzip.c + * + *------------------------------------------------------------------------- + */ +#include "postgres_fe.h" +#include <unistd.h> + +#include "compress_gzip.h" +#include "pg_backup_utils.h" + +#ifdef HAVE_LIBZ +#include "zlib.h" + +/*---------------------- + * Compressor API + *---------------------- + */ +typedef struct GzipCompressorState +{ + z_streamp zp; + + void *outbuf; + size_t outsize; +} GzipCompressorState; + +/* Private routines that support gzip compressed data I/O */ +static void +DeflateCompressorGzip(ArchiveHandle *AH, CompressorState *cs, bool flush) +{ + GzipCompressorState *gzipcs = (GzipCompressorState *) cs->private_data; + z_streamp zp = gzipcs->zp; + void *out = gzipcs->outbuf; + int res = Z_OK; + + while (gzipcs->zp->avail_in != 0 || flush) + { + res = deflate(zp, flush ? Z_FINISH : Z_NO_FLUSH); + if (res == Z_STREAM_ERROR) + pg_fatal("could not compress data: %s", zp->msg); + if ((flush && (zp->avail_out < gzipcs->outsize)) + || (zp->avail_out == 0) + || (zp->avail_in != 0) + ) + { + /* + * Extra paranoia: avoid zero-length chunks, since a zero length + * chunk is the EOF marker in the custom format. This should never + * happen but... + */ + if (zp->avail_out < gzipcs->outsize) + { + /* + * Any write function should do its own error checking but to + * make sure we do a check here as well... + */ + size_t len = gzipcs->outsize - zp->avail_out; + + cs->writeF(AH, (char *) out, len); + } + zp->next_out = out; + zp->avail_out = gzipcs->outsize; + } + + if (res == Z_STREAM_END) + break; + } +} + +static void +EndCompressorGzip(ArchiveHandle *AH, CompressorState *cs) +{ + GzipCompressorState *gzipcs = (GzipCompressorState *) cs->private_data; + z_streamp zp; + + if (gzipcs->zp) + { + zp = gzipcs->zp; + zp->next_in = NULL; + zp->avail_in = 0; + + /* Flush any remaining data from zlib buffer */ + DeflateCompressorGzip(AH, cs, true); + + if (deflateEnd(zp) != Z_OK) + pg_fatal("could not close compression stream: %s", zp->msg); + + pg_free(gzipcs->outbuf); + pg_free(gzipcs->zp); + } + + pg_free(gzipcs); + cs->private_data = NULL; +} + +static void +WriteDataToArchiveGzip(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen) +{ + GzipCompressorState *gzipcs = (GzipCompressorState *) cs->private_data; + z_streamp zp; + + if (!gzipcs->zp) + { + zp = gzipcs->zp = (z_streamp) pg_malloc(sizeof(z_stream)); + zp->zalloc = Z_NULL; + zp->zfree = Z_NULL; + zp->opaque = Z_NULL; + + /* + * outsize is the buffer size we tell zlib it can output to. We + * actually allocate one extra byte because some routines want to + * append a trailing zero byte to the zlib output. + */ + gzipcs->outbuf = pg_malloc(ZLIB_OUT_SIZE + 1); + gzipcs->outsize = ZLIB_OUT_SIZE; + + /* + * A level of zero simply copies the input one block at the time. This + * is probably not what the user wanted when calling this interface. + */ + if (cs->compression_spec.level == 0) + pg_fatal("requested to compress the archive yet no level was specified"); + + if (deflateInit(zp, cs->compression_spec.level) != Z_OK) + pg_fatal("could not initialize compression library: %s", zp->msg); + + /* Just be paranoid - maybe End is called after Start, with no Write */ + zp->next_out = gzipcs->outbuf; + zp->avail_out = gzipcs->outsize; + } + + gzipcs->zp->next_in = (void *) unconstify(void *, data); + gzipcs->zp->avail_in = dLen; + DeflateCompressorGzip(AH, cs, false); +} + +static void +ReadDataFromArchiveGzip(ArchiveHandle *AH, CompressorState *cs) +{ + z_streamp zp; + char *out; + int res = Z_OK; + size_t cnt; + char *buf; + size_t buflen; + + zp = (z_streamp) pg_malloc(sizeof(z_stream)); + zp->zalloc = Z_NULL; + zp->zfree = Z_NULL; + zp->opaque = Z_NULL; + + buf = pg_malloc(ZLIB_IN_SIZE); + buflen = ZLIB_IN_SIZE; + + out = pg_malloc(ZLIB_OUT_SIZE + 1); + + if (inflateInit(zp) != Z_OK) + pg_fatal("could not initialize compression library: %s", + zp->msg); + + /* no minimal chunk size for zlib */ + while ((cnt = cs->readF(AH, &buf, &buflen))) + { + zp->next_in = (void *) buf; + zp->avail_in = cnt; + + while (zp->avail_in > 0) + { + zp->next_out = (void *) out; + zp->avail_out = ZLIB_OUT_SIZE; + + res = inflate(zp, 0); + if (res != Z_OK && res != Z_STREAM_END) + pg_fatal("could not uncompress data: %s", zp->msg); + + out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; + ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); + } + } + + zp->next_in = NULL; + zp->avail_in = 0; + while (res != Z_STREAM_END) + { + zp->next_out = (void *) out; + zp->avail_out = ZLIB_OUT_SIZE; + res = inflate(zp, 0); + if (res != Z_OK && res != Z_STREAM_END) + pg_fatal("could not uncompress data: %s", zp->msg); + + out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; + ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); + } + + if (inflateEnd(zp) != Z_OK) + pg_fatal("could not close compression library: %s", zp->msg); + + free(buf); + free(out); + free(zp); +} + +/* Public routines that support gzip compressed data I/O */ +void +InitCompressorGzip(CompressorState *cs, const pg_compress_specification compression_spec) +{ + GzipCompressorState *gzipcs; + + cs->readData = ReadDataFromArchiveGzip; + cs->writeData = WriteDataToArchiveGzip; + cs->end = EndCompressorGzip; + + cs->compression_spec = compression_spec; + + gzipcs = (GzipCompressorState *) pg_malloc0(sizeof(GzipCompressorState)); + + cs->private_data = gzipcs; +} + + +/*---------------------- + * Compress File API + *---------------------- + */ + +static size_t +Gzip_read(void *ptr, size_t size, CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + size_t ret; + + ret = gzread(gzfp, ptr, size); + if (ret != size && !gzeof(gzfp)) + { + int errnum; + const char *errmsg = gzerror(gzfp, &errnum); + + pg_fatal("could not read from input file: %s", + errnum == Z_ERRNO ? strerror(errno) : errmsg); + } + + return ret; +} + +static size_t +Gzip_write(const void *ptr, size_t size, CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + + return gzwrite(gzfp, ptr, size); +} + +static int +Gzip_getc(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + int ret; + + errno = 0; + ret = gzgetc(gzfp); + if (ret == EOF) + { + if (!gzeof(gzfp)) + pg_fatal("could not read from input file: %s", strerror(errno)); + else + pg_fatal("could not read from input file: end of file"); + } + + return ret; +} + +static char * +Gzip_gets(char *ptr, int size, CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + + return gzgets(gzfp, ptr, size); +} + +static int +Gzip_close(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + int save_errno; + int ret; + + CFH->private_data = NULL; + + ret = gzclose(gzfp); + + save_errno = errno; + errno = save_errno; + + return ret; +} + +static int +Gzip_eof(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + + return gzeof(gzfp); +} + +static const char * +Gzip_get_error(CompressFileHandle *CFH) +{ + gzFile gzfp = (gzFile) CFH->private_data; + const char *errmsg; + int errnum; + + errmsg = gzerror(gzfp, &errnum); + if (errnum == Z_ERRNO) + errmsg = strerror(errno); + + return errmsg; +} + +static int +Gzip_open(const char *path, int fd, const char *mode, CompressFileHandle *CFH) +{ + gzFile gzfp; + char mode_compression[32]; + + if (CFH->compression_spec.level != Z_DEFAULT_COMPRESSION) + { + /* + * user has specified a compression level, so tell zlib to use it + */ + snprintf(mode_compression, sizeof(mode_compression), "%s%d", + mode, CFH->compression_spec.level); + } + else + strcpy(mode_compression, mode); + + if (fd >= 0) + gzfp = gzdopen(dup(fd), mode_compression); + else + gzfp = gzopen(path, mode_compression); + + if (gzfp == NULL) + return 1; + + CFH->private_data = gzfp; + + return 0; +} + +static int +Gzip_open_write(const char *path, const char *mode, CompressFileHandle *CFH) +{ + char *fname; + int ret; + int save_errno; + + fname = psprintf("%s.gz", path); + ret = CFH->open_func(fname, -1, mode, CFH); + + save_errno = errno; + pg_free(fname); + errno = save_errno; + + return ret; +} + +void +InitCompressGzip(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +{ + CFH->open_func = Gzip_open; + CFH->open_write_func = Gzip_open_write; + CFH->read_func = Gzip_read; + CFH->write_func = Gzip_write; + CFH->gets_func = Gzip_gets; + CFH->getc_func = Gzip_getc; + CFH->close_func = Gzip_close; + CFH->eof_func = Gzip_eof; + CFH->get_error_func = Gzip_get_error; + + CFH->compression_spec = compression_spec; + + CFH->private_data = NULL; +} +#else /* HAVE_LIBZ */ +void +InitCompressorGzip(CompressorState *cs, const pg_compress_specification compression_spec) +{ + pg_fatal("this build does not support compression with %s", "gzip"); +} + +void +InitCompressGzip(CompressFileHandle *CFH, const pg_compress_specification compression_spec) +{ + pg_fatal("this build does not support compression with %s", "gzip"); +} +#endif /* HAVE_LIBZ */ diff --git a/src/bin/pg_dump/compress_gzip.h b/src/bin/pg_dump/compress_gzip.h new file mode 100644 index 0000000000..6dfd0eb04d --- /dev/null +++ b/src/bin/pg_dump/compress_gzip.h @@ -0,0 +1,22 @@ +/*------------------------------------------------------------------------- + * + * compress_gzip.h + * Interface to compress_io.c routines + * + * Portions Copyright (c) 1996-2022, PostgreSQL Global Development Group + * Portions Copyright (c) 1994, Regents of the University of California + * + * IDENTIFICATION + * src/bin/pg_dump/compress_gzip.h + * + *------------------------------------------------------------------------- + */ +#ifndef _COMPRESS_GZIP_H_ +#define _COMPRESS_GZIP_H_ + +#include "compress_io.h" + +extern void InitCompressorGzip(CompressorState *cs, const pg_compress_specification compression_spec); +extern void InitCompressGzip(CompressFileHandle *CFH, const pg_compress_specification compression_spec); + +#endif /* _COMPRESS_GZIP_H_ */ diff --git a/src/bin/pg_dump/compress_io.c b/src/bin/pg_dump/compress_io.c index 1db973b6a2..7274039e87 100644 --- a/src/bin/pg_dump/compress_io.c +++ b/src/bin/pg_dump/compress_io.c @@ -9,42 +9,44 @@ * * This file includes two APIs for dealing with compressed data. The first * provides more flexibility, using callbacks to read/write data from the - * underlying stream. The second API is a wrapper around fopen/gzopen and + * underlying stream. The second API is a wrapper around fopen and * friends, providing an interface similar to those, but abstracts away - * the possible compression. Both APIs use libz for the compression, but - * the second API uses gzip headers, so the resulting files can be easily - * manipulated with the gzip utility. + * the possible compression. The second API is aimed for the resulting + * files can be easily manipulated with an external compression utility + * program. * * Compressor API * -------------- * * The interface for writing to an archive consists of three functions: - * AllocateCompressor, WriteDataToArchive and EndCompressor. First you call - * AllocateCompressor, then write all the data by calling WriteDataToArchive - * as many times as needed, and finally EndCompressor. WriteDataToArchive - * and EndCompressor will call the WriteFunc that was provided to - * AllocateCompressor for each chunk of compressed data. + * AllocateCompressor, writeData, and EndCompressor. First you call + * AllocateCompressor, then write all the data by calling writeData as many + * times as needed, and finally EndCompressor. writeData will call the + * WriteFunc that was provided to AllocateCompressor for each chunk of + * compressed data. * - * The interface for reading an archive consists of just one function: - * ReadDataFromArchive. ReadDataFromArchive reads the whole compressed input - * stream, by repeatedly calling the given ReadFunc. ReadFunc returns the - * compressed data chunk at a time, and ReadDataFromArchive decompresses it - * and passes the decompressed data to ahwrite(), until ReadFunc returns 0 - * to signal EOF. - * - * The interface is the same for compressed and uncompressed streams. + * The interface for reading an archive consists of the same three functions: + * AllocateCompressor, readData, and EndCompressor. First you call + * AllocateCompressor, then read all the data by calling readData to read the + * whole compressed stream which repeatedly calls the given ReadFunc. ReadFunc + * returns the compressed data chunk at a time, and readData decompresses it + * and passes the decompressed data to ahwrite(), until ReadFunc returns 0 to + * signal EOF. The interface is the same for compressed and uncompressed + * streams. * * Compressed stream API * ---------------------- * * The compressed stream API is a wrapper around the C standard fopen() and - * libz's gzopen() APIs. It allows you to use the same functions for - * compressed and uncompressed streams. cfopen_read() first tries to open - * the file with given name, and if it fails, it tries to open the same - * file with the .gz suffix. cfopen_write() opens a file for writing, an - * extra argument specifies if the file should be compressed, and adds the - * .gz suffix to the filename if so. This allows you to easily handle both - * compressed and uncompressed files. + * libz's gzopen() APIs and custom LZ4 calls which provide similar + * functionality. It allows you to use the same functions for compressed and + * uncompressed streams. cfopen_read() first tries to open the file with given + * name, and if it fails, it tries to open the same file with the .gz suffix, + * failing that it tries to open the same file with the .lz4 suffix. + * cfopen_write() opens a file for writing, an extra argument specifies the + * method to use should the file be compressed, and adds the appropriate + * suffix, .gz or .lz4, to the filename if so. This allows you to easily handle + * both compressed and uncompressed files. * * IDENTIFICATION * src/bin/pg_dump/compress_io.c @@ -53,7 +55,11 @@ */ #include "postgres_fe.h" +#include <sys/stat.h> +#include <unistd.h> + #include "compress_io.h" +#include "compress_gzip.h" #include "pg_backup_utils.h" #ifdef HAVE_LIBZ @@ -65,84 +71,69 @@ *---------------------- */ -/* typedef appears in compress_io.h */ -struct CompressorState +static void +ReadDataFromArchiveNone(ArchiveHandle *AH, CompressorState *cs) { - pg_compress_specification compression_spec; - WriteFunc writeF; + size_t cnt; + char *buf; + size_t buflen; -#ifdef HAVE_LIBZ - z_streamp zp; - char *zlibOut; - size_t zlibOutSize; -#endif -}; + buf = pg_malloc(ZLIB_OUT_SIZE); + buflen = ZLIB_OUT_SIZE; + + while ((cnt = cs->readF(AH, &buf, &buflen))) + { + ahwrite(buf, 1, cnt, AH); + } + + free(buf); +} -/* Routines that support zlib compressed data I/O */ -#ifdef HAVE_LIBZ -static void InitCompressorZlib(CompressorState *cs, int level); -static void DeflateCompressorZlib(ArchiveHandle *AH, CompressorState *cs, - bool flush); -static void ReadDataFromArchiveZlib(ArchiveHandle *AH, ReadFunc readF); -static void WriteDataToArchiveZlib(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen); -static void EndCompressorZlib(ArchiveHandle *AH, CompressorState *cs); -#endif -/* Routines that support uncompressed data I/O */ -static void ReadDataFromArchiveNone(ArchiveHandle *AH, ReadFunc readF); -static void WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen); +static void +WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen) +{ + cs->writeF(AH, data, dLen); +} + +static void +EndCompressorNone(ArchiveHandle *AH, CompressorState *cs) +{ + /* no op */ +} + +static void +InitCompressorNone(CompressorState *cs, + const pg_compress_specification compression_spec) +{ + cs->readData = ReadDataFromArchiveNone; + cs->writeData = WriteDataToArchiveNone; + cs->end = EndCompressorNone; + + cs->compression_spec = compression_spec; +} /* Public interface routines */ /* Allocate a new compressor */ CompressorState * AllocateCompressor(const pg_compress_specification compression_spec, - WriteFunc writeF) + ReadFunc readF, WriteFunc writeF) { CompressorState *cs; -#ifndef HAVE_LIBZ - if (compression_spec.algorithm == PG_COMPRESSION_GZIP) - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - cs = (CompressorState *) pg_malloc0(sizeof(CompressorState)); + cs->readF = readF; cs->writeF = writeF; - cs->compression_spec = compression_spec; - /* - * Perform compression algorithm specific initialization. - */ -#ifdef HAVE_LIBZ - if (cs->compression_spec.algorithm == PG_COMPRESSION_GZIP) - InitCompressorZlib(cs, cs->compression_spec.level); -#endif - - return cs; -} - -/* - * Read all compressed data from the input stream (via readF) and print it - * out with ahwrite(). - */ -void -ReadDataFromArchive(ArchiveHandle *AH, - const pg_compress_specification compression_spec, - ReadFunc readF) -{ switch (compression_spec.algorithm) { case PG_COMPRESSION_NONE: - ReadDataFromArchiveNone(AH, readF); + InitCompressorNone(cs, compression_spec); break; case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ReadDataFromArchiveZlib(AH, readF); -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif + InitCompressorGzip(cs, compression_spec); break; case PG_COMPRESSION_LZ4: pg_fatal("compression with %s is not yet supported", "LZ4"); @@ -151,34 +142,8 @@ ReadDataFromArchive(ArchiveHandle *AH, pg_fatal("compression with %s is not yet supported", "ZSTD"); break; } -} -/* - * Compress and write data to the output stream (via writeF). - */ -void -WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, - const void *data, size_t dLen) -{ - switch (cs->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - WriteDataToArchiveNone(AH, cs, data, dLen); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - WriteDataToArchiveZlib(AH, cs, data, dLen); -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } + return cs; } /* @@ -187,401 +152,178 @@ WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, void EndCompressor(ArchiveHandle *AH, CompressorState *cs) { - switch (cs->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - EndCompressorZlib(AH, cs); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - free(cs); + cs->end(AH, cs); + pg_free(cs); } -/* Private routines, specific to each compression method. */ - -#ifdef HAVE_LIBZ -/* - * Functions for zlib compressed output. +/*---------------------- + * Compressed stream API + *---------------------- */ -static void -InitCompressorZlib(CompressorState *cs, int level) +static int +hasSuffix(const char *filename, const char *suffix) { - z_streamp zp; - - zp = cs->zp = (z_streamp) pg_malloc(sizeof(z_stream)); - zp->zalloc = Z_NULL; - zp->zfree = Z_NULL; - zp->opaque = Z_NULL; - - /* - * zlibOutSize is the buffer size we tell zlib it can output to. We - * actually allocate one extra byte because some routines want to append a - * trailing zero byte to the zlib output. - */ - cs->zlibOut = (char *) pg_malloc(ZLIB_OUT_SIZE + 1); - cs->zlibOutSize = ZLIB_OUT_SIZE; - - if (deflateInit(zp, level) != Z_OK) - pg_fatal("could not initialize compression library: %s", - zp->msg); - - /* Just be paranoid - maybe End is called after Start, with no Write */ - zp->next_out = (void *) cs->zlibOut; - zp->avail_out = cs->zlibOutSize; + int filenamelen = strlen(filename); + int suffixlen = strlen(suffix); + + if (filenamelen < suffixlen) + return 0; + + return memcmp(&filename[filenamelen - suffixlen], + suffix, + suffixlen) == 0; } +/* free() without changing errno; useful in several places below */ static void -EndCompressorZlib(ArchiveHandle *AH, CompressorState *cs) +free_keep_errno(void *p) { - z_streamp zp = cs->zp; - - zp->next_in = NULL; - zp->avail_in = 0; - - /* Flush any remaining data from zlib buffer */ - DeflateCompressorZlib(AH, cs, true); - - if (deflateEnd(zp) != Z_OK) - pg_fatal("could not close compression stream: %s", zp->msg); + int save_errno = errno; - free(cs->zlibOut); - free(cs->zp); + free(p); + errno = save_errno; } -static void -DeflateCompressorZlib(ArchiveHandle *AH, CompressorState *cs, bool flush) +/* + * Compression None implementation + */ +static size_t +read_none(void *ptr, size_t size, CompressFileHandle *CFH) { - z_streamp zp = cs->zp; - char *out = cs->zlibOut; - int res = Z_OK; + FILE *fp = (FILE *) CFH->private_data; + size_t ret; - while (cs->zp->avail_in != 0 || flush) - { - res = deflate(zp, flush ? Z_FINISH : Z_NO_FLUSH); - if (res == Z_STREAM_ERROR) - pg_fatal("could not compress data: %s", zp->msg); - if ((flush && (zp->avail_out < cs->zlibOutSize)) - || (zp->avail_out == 0) - || (zp->avail_in != 0) - ) - { - /* - * Extra paranoia: avoid zero-length chunks, since a zero length - * chunk is the EOF marker in the custom format. This should never - * happen but... - */ - if (zp->avail_out < cs->zlibOutSize) - { - /* - * Any write function should do its own error checking but to - * make sure we do a check here as well... - */ - size_t len = cs->zlibOutSize - zp->avail_out; - - cs->writeF(AH, out, len); - } - zp->next_out = (void *) out; - zp->avail_out = cs->zlibOutSize; - } + if (size == 0) + return 0; - if (res == Z_STREAM_END) - break; - } + ret = fread(ptr, 1, size, fp); + if (ret != size && !feof(fp)) + pg_fatal("could not read from input file: %s", + strerror(errno)); + + return ret; } -static void -WriteDataToArchiveZlib(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen) +static size_t +write_none(const void *ptr, size_t size, CompressFileHandle *CFH) { - cs->zp->next_in = (void *) unconstify(char *, data); - cs->zp->avail_in = dLen; - DeflateCompressorZlib(AH, cs, false); + return fwrite(ptr, 1, size, (FILE *) CFH->private_data); } -static void -ReadDataFromArchiveZlib(ArchiveHandle *AH, ReadFunc readF) +static const char * +get_error_none(CompressFileHandle *CFH) { - z_streamp zp; - char *out; - int res = Z_OK; - size_t cnt; - char *buf; - size_t buflen; - - zp = (z_streamp) pg_malloc(sizeof(z_stream)); - zp->zalloc = Z_NULL; - zp->zfree = Z_NULL; - zp->opaque = Z_NULL; - - buf = pg_malloc(ZLIB_IN_SIZE); - buflen = ZLIB_IN_SIZE; - - out = pg_malloc(ZLIB_OUT_SIZE + 1); - - if (inflateInit(zp) != Z_OK) - pg_fatal("could not initialize compression library: %s", - zp->msg); - - /* no minimal chunk size for zlib */ - while ((cnt = readF(AH, &buf, &buflen))) - { - zp->next_in = (void *) buf; - zp->avail_in = cnt; - - while (zp->avail_in > 0) - { - zp->next_out = (void *) out; - zp->avail_out = ZLIB_OUT_SIZE; + return strerror(errno); +} - res = inflate(zp, 0); - if (res != Z_OK && res != Z_STREAM_END) - pg_fatal("could not uncompress data: %s", zp->msg); +static char * +gets_none(char *ptr, int size, CompressFileHandle *CFH) +{ + return fgets(ptr, size, (FILE *) CFH->private_data); +} - out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; - ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); - } - } +static int +getc_none(CompressFileHandle *CFH) +{ + FILE *fp = (FILE *) CFH->private_data; + int ret; - zp->next_in = NULL; - zp->avail_in = 0; - while (res != Z_STREAM_END) + ret = fgetc(fp); + if (ret == EOF) { - zp->next_out = (void *) out; - zp->avail_out = ZLIB_OUT_SIZE; - res = inflate(zp, 0); - if (res != Z_OK && res != Z_STREAM_END) - pg_fatal("could not uncompress data: %s", zp->msg); - - out[ZLIB_OUT_SIZE - zp->avail_out] = '\0'; - ahwrite(out, 1, ZLIB_OUT_SIZE - zp->avail_out, AH); + if (!feof(fp)) + pg_fatal("could not read from input file: %s", strerror(errno)); + else + pg_fatal("could not read from input file: end of file"); } - if (inflateEnd(zp) != Z_OK) - pg_fatal("could not close compression library: %s", zp->msg); - - free(buf); - free(out); - free(zp); + return ret; } -#endif /* HAVE_LIBZ */ - -/* - * Functions for uncompressed output. - */ - -static void -ReadDataFromArchiveNone(ArchiveHandle *AH, ReadFunc readF) +static int +close_none(CompressFileHandle *CFH) { - size_t cnt; - char *buf; - size_t buflen; + FILE *fp = (FILE *) CFH->private_data; + int ret = 0; - buf = pg_malloc(ZLIB_OUT_SIZE); - buflen = ZLIB_OUT_SIZE; + CFH->private_data = NULL; - while ((cnt = readF(AH, &buf, &buflen))) - { - ahwrite(buf, 1, cnt, AH); - } + if (fp) + ret = fclose(fp); - free(buf); + return ret; } -static void -WriteDataToArchiveNone(ArchiveHandle *AH, CompressorState *cs, - const char *data, size_t dLen) + +static int +eof_none(CompressFileHandle *CFH) { - cs->writeF(AH, data, dLen); + return feof((FILE *) CFH->private_data); } - -/*---------------------- - * Compressed stream API - *---------------------- - */ - -/* - * cfp represents an open stream, wrapping the underlying FILE or gzFile - * pointer. This is opaque to the callers. - */ -struct cfp +static int +open_none(const char *path, int fd, const char *mode, CompressFileHandle *CFH) { - pg_compress_specification compression_spec; - void *fp; -}; + Assert(CFH->private_data == NULL); -#ifdef HAVE_LIBZ -static int hasSuffix(const char *filename, const char *suffix); -#endif + if (fd >= 0) + CFH->private_data = fdopen(dup(fd), mode); + else + CFH->private_data = fopen(path, mode); -/* free() without changing errno; useful in several places below */ -static void -free_keep_errno(void *p) -{ - int save_errno = errno; + if (CFH->private_data == NULL) + return 1; - free(p); - errno = save_errno; + return 0; } -/* - * Open a file for reading. 'path' is the file to open, and 'mode' should - * be either "r" or "rb". - * - * If the file at 'path' does not exist, we append the ".gz" suffix (if 'path' - * doesn't already have it) and try again. So if you pass "foo" as 'path', - * this will open either "foo" or "foo.gz". - * - * On failure, return NULL with an error code in errno. - */ -cfp * -cfopen_read(const char *path, const char *mode) +static int +open_write_none(const char *path, const char *mode, CompressFileHandle *CFH) { - cfp *fp; - - pg_compress_specification compression_spec = {0}; + Assert(CFH->private_data == NULL); -#ifdef HAVE_LIBZ - if (hasSuffix(path, ".gz")) - { - compression_spec.algorithm = PG_COMPRESSION_GZIP; - fp = cfopen(path, mode, compression_spec); - } - else -#endif - { - compression_spec.algorithm = PG_COMPRESSION_NONE; - fp = cfopen(path, mode, compression_spec); -#ifdef HAVE_LIBZ - if (fp == NULL) - { - char *fname; + CFH->private_data = fopen(path, mode); + if (CFH->private_data == NULL) + return 1; - fname = psprintf("%s.gz", path); - compression_spec.algorithm = PG_COMPRESSION_GZIP; - fp = cfopen(fname, mode, compression_spec); - free_keep_errno(fname); - } -#endif - } - return fp; + return 0; } -/* - * Open a file for writing. 'path' indicates the path name, and 'mode' must - * be a filemode as accepted by fopen() and gzopen() that indicates writing - * ("w", "wb", "a", or "ab"). - * - * If 'compression_spec.algorithm' is GZIP, a gzip compressed stream is opened, - * and 'compression_spec.level' used. The ".gz" suffix is automatically added to - * 'path' in that case. - * - * On failure, return NULL with an error code in errno. - */ -cfp * -cfopen_write(const char *path, const char *mode, - const pg_compress_specification compression_spec) +static void +InitCompressNone(CompressFileHandle *CFH, + const pg_compress_specification compression_spec) { - cfp *fp; - - if (compression_spec.algorithm == PG_COMPRESSION_NONE) - fp = cfopen(path, mode, compression_spec); - else - { -#ifdef HAVE_LIBZ - char *fname; - - fname = psprintf("%s.gz", path); - fp = cfopen(fname, mode, compression_spec); - free_keep_errno(fname); -#else - pg_fatal("this build does not support compression with %s", "gzip"); - fp = NULL; /* keep compiler quiet */ -#endif - } - return fp; + CFH->open_func = open_none; + CFH->open_write_func = open_write_none; + CFH->read_func = read_none; + CFH->write_func = write_none; + CFH->gets_func = gets_none; + CFH->getc_func = getc_none; + CFH->close_func = close_none; + CFH->eof_func = eof_none; + CFH->get_error_func = get_error_none; + + CFH->private_data = NULL; } /* - * This is the workhorse for cfopen() or cfdopen(). It opens file 'path' or - * associates a stream 'fd', if 'fd' is a valid descriptor, in 'mode'. The - * descriptor is not dup'ed and it is the caller's responsibility to do so. - * The caller must verify that the 'compress_algorithm' is supported by the - * current build. - * - * On failure, return NULL with an error code in errno. + * Public interface */ -static cfp * -cfopen_internal(const char *path, int fd, const char *mode, - pg_compress_specification compression_spec) +CompressFileHandle * +InitCompressFileHandle(const pg_compress_specification compression_spec) { - cfp *fp = pg_malloc(sizeof(cfp)); + CompressFileHandle *CFH; - fp->compression_spec = compression_spec; + CFH = pg_malloc0(sizeof(CompressFileHandle)); switch (compression_spec.algorithm) { case PG_COMPRESSION_NONE: - if (fd >= 0) - fp->fp = fdopen(fd, mode); - else - fp->fp = fopen(path, mode); - if (fp->fp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } - + InitCompressNone(CFH, compression_spec); break; case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - if (compression_spec.level != Z_DEFAULT_COMPRESSION) - { - /* - * user has specified a compression level, so tell zlib to use - * it - */ - char mode_compression[32]; - - snprintf(mode_compression, sizeof(mode_compression), "%s%d", - mode, compression_spec.level); - if (fd >= 0) - fp->fp = gzdopen(fd, mode_compression); - else - fp->fp = gzopen(path, mode_compression); - } - else - { - /* don't specify a level, just use the zlib default */ - if (fd >= 0) - fp->fp = gzdopen(fd, mode); - else - fp->fp = gzopen(path, mode); - } - - if (fp->fp == NULL) - { - free_keep_errno(fp); - fp = NULL; - } -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif + InitCompressGzip(CFH, compression_spec); break; case PG_COMPRESSION_LZ4: pg_fatal("compression with %s is not yet supported", "LZ4"); @@ -591,266 +333,88 @@ cfopen_internal(const char *path, int fd, const char *mode, break; } - return fp; + return CFH; } -cfp * -cfopen(const char *path, const char *mode, - const pg_compress_specification compression_spec) +/* + * Open a file for reading. 'path' is the file to open, and 'mode' should + * be either "r" or "rb". + * + * If the file at 'path' does not exist, we append the "{.gz,.lz4}" suffix (i + * 'path' doesn't already have it) and try again. So if you pass "foo" as + * 'path', this will open either "foo" or "foo.gz" or "foo.lz4", trying in that + * order. + * + * On failure, return NULL with an error code in errno. + */ +CompressFileHandle * +InitDiscoverCompressFileHandle(const char *path, const char *mode) { - return cfopen_internal(path, -1, mode, compression_spec); -} + CompressFileHandle *CFH = NULL; + struct stat st; + char *fname; + pg_compress_specification compression_spec = {0}; -cfp * -cfdopen(int fd, const char *mode, - const pg_compress_specification compression_spec) -{ - return cfopen_internal(NULL, fd, mode, compression_spec); -} + compression_spec.algorithm = PG_COMPRESSION_NONE; -int -cfread(void *ptr, int size, cfp *fp) -{ - int ret = 0; + Assert(strcmp(mode, "r") == 0 || strcmp(mode, "rb") == 0); - if (size == 0) - return 0; + fname = strdup(path); - switch (fp->compression_spec.algorithm) + if (hasSuffix(fname, ".gz")) + compression_spec.algorithm = PG_COMPRESSION_GZIP; + else { - case PG_COMPRESSION_NONE: - ret = fread(ptr, 1, size, (FILE *) fp->fp); - if (ret != size && !feof((FILE *) fp->fp)) - READ_ERROR_EXIT((FILE *) fp->fp); + bool exists; - break; - case PG_COMPRESSION_GZIP: + exists = (stat(path, &st) == 0); + /* avoid unused warning if it is not build with compression */ + if (exists) + compression_spec.algorithm = PG_COMPRESSION_NONE; #ifdef HAVE_LIBZ - ret = gzread((gzFile) fp->fp, ptr, size); - if (ret != size && !gzeof((gzFile) fp->fp)) - { - int errnum; - const char *errmsg = gzerror((gzFile) fp->fp, &errnum); - - pg_fatal("could not read from input file: %s", - errnum == Z_ERRNO ? strerror(errno) : errmsg); - } -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - return ret; -} - -int -cfwrite(const void *ptr, int size, cfp *fp) -{ - int ret = 0; - - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = fwrite(ptr, 1, size, (FILE *) fp->fp); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzwrite((gzFile) fp->fp, ptr, size); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - return ret; -} - -int -cfgetc(cfp *fp) -{ - int ret = 0; - - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = fgetc((FILE *) fp->fp); - if (ret == EOF) - READ_ERROR_EXIT((FILE *) fp->fp); + if (!exists) + { + free_keep_errno(fname); + fname = psprintf("%s.gz", path); + exists = (stat(fname, &st) == 0); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzgetc((gzFile) fp->fp); - if (ret == EOF) - { - if (!gzeof((gzFile) fp->fp)) - pg_fatal("could not read from input file: %s", strerror(errno)); - else - pg_fatal("could not read from input file: end of file"); - } -#else - pg_fatal("this build does not support compression with %s", - "gzip"); + if (exists) + compression_spec.algorithm = PG_COMPRESSION_GZIP; + } #endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - return ret; -} - -char * -cfgets(cfp *fp, char *buf, int len) -{ - char *ret = NULL; - - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = fgets(buf, len, (FILE *) fp->fp); +#ifdef USE_LZ4 + if (!exists) + { + free_keep_errno(fname); + fname = psprintf("%s.lz4", path); + exists = (stat(fname, &st) == 0); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzgets((gzFile) fp->fp, buf, len); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); + if (exists) + compression_spec.algorithm = PG_COMPRESSION_LZ4; + } #endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } - - return ret; -} - -int -cfclose(cfp *fp) -{ - int ret = 0; - - if (fp == NULL) - { - errno = EBADF; - return EOF; } - switch (fp->compression_spec.algorithm) + CFH = InitCompressFileHandle(compression_spec); + if (CFH->open_func(fname, -1, mode, CFH)) { - case PG_COMPRESSION_NONE: - ret = fclose((FILE *) fp->fp); - fp->fp = NULL; - - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzclose((gzFile) fp->fp); - fp->fp = NULL; -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; + free_keep_errno(CFH); + CFH = NULL; } + free_keep_errno(fname); - free_keep_errno(fp); - - return ret; + return CFH; } int -cfeof(cfp *fp) +DestroyCompressFileHandle(CompressFileHandle *CFH) { int ret = 0; - switch (fp->compression_spec.algorithm) - { - case PG_COMPRESSION_NONE: - ret = feof((FILE *) fp->fp); + if (CFH->private_data) + ret = CFH->close_func(CFH); - break; - case PG_COMPRESSION_GZIP: -#ifdef HAVE_LIBZ - ret = gzeof((gzFile) fp->fp); -#else - pg_fatal("this build does not support compression with %s", - "gzip"); -#endif - break; - case PG_COMPRESSION_LZ4: - pg_fatal("compression with %s is not yet supported", "LZ4"); - break; - case PG_COMPRESSION_ZSTD: - pg_fatal("compression with %s is not yet supported", "ZSTD"); - break; - } + free_keep_errno(CFH); return ret; } - -const char * -get_cfp_error(cfp *fp) -{ - if (fp->compression_spec.algorithm == PG_COMPRESSION_GZIP) - { -#ifdef HAVE_LIBZ - int errnum; - const char *errmsg = gzerror((gzFile) fp->fp, &errnum); - - if (errnum != Z_ERRNO) - return errmsg; -#else - pg_fatal("this build does not support compression with %s", "gzip"); -#endif - } - - return strerror(errno); -} - -#ifdef HAVE_LIBZ -static int -hasSuffix(const char *filename, const char *suffix) -{ - int filenamelen = strlen(filename); - int suffixlen = strlen(suffix); - - if (filenamelen < suffixlen) - return 0; - - return memcmp(&filename[filenamelen - suffixlen], - suffix, - suffixlen) == 0; -} - -#endif diff --git a/src/bin/pg_dump/compress_io.h b/src/bin/pg_dump/compress_io.h index a429dc4789..4b4a00c010 100644 --- a/src/bin/pg_dump/compress_io.h +++ b/src/bin/pg_dump/compress_io.h @@ -37,32 +37,62 @@ typedef void (*WriteFunc) (ArchiveHandle *AH, const char *buf, size_t len); */ typedef size_t (*ReadFunc) (ArchiveHandle *AH, char **buf, size_t *buflen); -/* struct definition appears in compress_io.c */ typedef struct CompressorState CompressorState; +struct CompressorState +{ + /* + * Read all compressed data from the input stream (via readF) and print it + * out with ahwrite(). + */ + void (*readData) (ArchiveHandle *AH, CompressorState *cs); + + /* + * Compress and write data to the output stream (via writeF). + */ + void (*writeData) (ArchiveHandle *AH, CompressorState *cs, + const void *data, size_t dLen); + void (*end) (ArchiveHandle *AH, CompressorState *cs); + + ReadFunc readF; + WriteFunc writeF; + + pg_compress_specification compression_spec; + void *private_data; +}; extern CompressorState *AllocateCompressor(const pg_compress_specification compression_spec, + ReadFunc readF, WriteFunc writeF); -extern void ReadDataFromArchive(ArchiveHandle *AH, - const pg_compress_specification compression_spec, - ReadFunc readF); -extern void WriteDataToArchive(ArchiveHandle *AH, CompressorState *cs, - const void *data, size_t dLen); extern void EndCompressor(ArchiveHandle *AH, CompressorState *cs); +/* + * Compress File Handle + */ +typedef struct CompressFileHandle CompressFileHandle; + +struct CompressFileHandle +{ + int (*open_func) (const char *path, int fd, const char *mode, + CompressFileHandle *CFH); + int (*open_write_func) (const char *path, const char *mode, + CompressFileHandle *cxt); + size_t (*read_func) (void *ptr, size_t size, CompressFileHandle *CFH); + size_t (*write_func) (const void *ptr, size_t size, + struct CompressFileHandle *CFH); + char *(*gets_func) (char *s, int size, CompressFileHandle *CFH); + int (*getc_func) (CompressFileHandle *CFH); + int (*eof_func) (CompressFileHandle *CFH); + int (*close_func) (CompressFileHandle *CFH); + const char *(*get_error_func) (CompressFileHandle *CFH); + + pg_compress_specification compression_spec; + void *private_data; +}; -typedef struct cfp cfp; +extern CompressFileHandle *InitCompressFileHandle(const pg_compress_specification compression_spec); -extern cfp *cfopen(const char *path, const char *mode, - const pg_compress_specification compression_spec); -extern cfp *cfopen_read(const char *path, const char *mode); -extern cfp *cfopen_write(const char *path, const char *mode, - const pg_compress_specification compression_spec); -extern int cfread(void *ptr, int size, cfp *fp); -extern int cfwrite(const void *ptr, int size, cfp *fp); -extern int cfgetc(cfp *fp); -extern char *cfgets(cfp *fp, char *buf, int len); -extern int cfclose(cfp *fp); -extern int cfeof(cfp *fp); -extern const char *get_cfp_error(cfp *fp); +extern CompressFileHandle *InitDiscoverCompressFileHandle(const char *path, + const char *mode); +extern int DestroyCompressFileHandle(CompressFileHandle *CFH); #endif diff --git a/src/bin/pg_dump/meson.build b/src/bin/pg_dump/meson.build index ca62f9a374..aa2c91829c 100644 --- a/src/bin/pg_dump/meson.build +++ b/src/bin/pg_dump/meson.build @@ -2,6 +2,7 @@ pg_dump_common_sources = files( 'compress_io.c', + 'compress_gzip.c', 'dumputils.c', 'parallel.c', 'pg_backup_archiver.c', diff --git a/src/bin/pg_dump/pg_backup_archiver.c b/src/bin/pg_dump/pg_backup_archiver.c index fb94317ad9..1f207c6f4d 100644 --- a/src/bin/pg_dump/pg_backup_archiver.c +++ b/src/bin/pg_dump/pg_backup_archiver.c @@ -95,8 +95,8 @@ static void dump_lo_buf(ArchiveHandle *AH); static void dumpTimestamp(ArchiveHandle *AH, const char *msg, time_t tim); static void SetOutput(ArchiveHandle *AH, const char *filename, const pg_compress_specification compression_spec); -static cfp *SaveOutput(ArchiveHandle *AH); -static void RestoreOutput(ArchiveHandle *AH, cfp *savedOutput); +static CompressFileHandle *SaveOutput(ArchiveHandle *AH); +static void RestoreOutput(ArchiveHandle *AH, CompressFileHandle *savedOutput); static int restore_toc_entry(ArchiveHandle *AH, TocEntry *te, bool is_parallel); static void restore_toc_entries_prefork(ArchiveHandle *AH, @@ -272,7 +272,7 @@ CloseArchive(Archive *AHX) /* Close the output */ errno = 0; - res = cfclose(AH->OF); + res = DestroyCompressFileHandle(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); @@ -355,7 +355,7 @@ RestoreArchive(Archive *AHX) bool parallel_mode; bool supports_compression; TocEntry *te; - cfp *sav; + CompressFileHandle *sav; AH->stage = STAGE_INITIALIZING; @@ -1127,7 +1127,7 @@ PrintTOCSummary(Archive *AHX) TocEntry *te; pg_compress_specification out_compression_spec = {0}; teSection curSection; - cfp *sav; + CompressFileHandle *sav; const char *fmtName; char stamp_str[64]; @@ -1143,9 +1143,10 @@ PrintTOCSummary(Archive *AHX) strcpy(stamp_str, "[unknown]"); ahprintf(AH, ";\n; Archive created at %s\n", stamp_str); - ahprintf(AH, "; dbname: %s\n; TOC Entries: %d\n; Compression: %d\n", + ahprintf(AH, "; dbname: %s\n; TOC Entries: %d\n; Compression: %s\n", sanitize_line(AH->archdbname, false), - AH->tocCount, AH->compression_spec.level); + AH->tocCount, + get_compress_algorithm_name(AH->compression_spec.algorithm)); switch (AH->format) { @@ -1502,6 +1503,7 @@ static void SetOutput(ArchiveHandle *AH, const char *filename, const pg_compress_specification compression_spec) { + CompressFileHandle *CFH; const char *mode; int fn = -1; @@ -1524,33 +1526,32 @@ SetOutput(ArchiveHandle *AH, const char *filename, else mode = PG_BINARY_W; - if (fn >= 0) - AH->OF = cfdopen(dup(fn), mode, compression_spec); - else - AH->OF = cfopen(filename, mode, compression_spec); + CFH = InitCompressFileHandle(compression_spec); - if (!AH->OF) + if (CFH->open_func(filename, fn, mode, CFH)) { if (filename) pg_fatal("could not open output file \"%s\": %m", filename); else pg_fatal("could not open output file: %m"); } + + AH->OF = CFH; } -static cfp * +static CompressFileHandle * SaveOutput(ArchiveHandle *AH) { - return (cfp *) AH->OF; + return (CompressFileHandle *) AH->OF; } static void -RestoreOutput(ArchiveHandle *AH, cfp *savedOutput) +RestoreOutput(ArchiveHandle *AH, CompressFileHandle *savedOutput) { int res; errno = 0; - res = cfclose(AH->OF); + res = DestroyCompressFileHandle(AH->OF); if (res != 0) pg_fatal("could not close output file: %m"); @@ -1689,7 +1690,11 @@ ahwrite(const void *ptr, size_t size, size_t nmemb, ArchiveHandle *AH) else if (RestoringToDB(AH)) bytes_written = ExecuteSqlCommandBuf(&AH->public, (const char *) ptr, size * nmemb); else - bytes_written = cfwrite(ptr, size * nmemb, AH->OF); + { + CompressFileHandle *CFH = (CompressFileHandle *) AH->OF; + + bytes_written = CFH->write_func(ptr, size * nmemb, CFH); + } if (bytes_written != size * nmemb) WRITE_ERROR_EXIT; @@ -2031,6 +2036,18 @@ ReadStr(ArchiveHandle *AH) return buf; } +static bool +_fileExistsInDirectory(const char *dir, const char *filename) +{ + struct stat st; + char buf[MAXPGPATH]; + + if (snprintf(buf, MAXPGPATH, "%s/%s", dir, filename) >= MAXPGPATH) + pg_fatal("directory name too long: \"%s\"", dir); + + return (stat(buf, &st) == 0 && S_ISREG(st.st_mode)); +} + static int _discoverArchiveFormat(ArchiveHandle *AH) { @@ -2061,26 +2078,12 @@ _discoverArchiveFormat(ArchiveHandle *AH) */ if (stat(AH->fSpec, &st) == 0 && S_ISDIR(st.st_mode)) { - char buf[MAXPGPATH]; - - if (snprintf(buf, MAXPGPATH, "%s/toc.dat", AH->fSpec) >= MAXPGPATH) - pg_fatal("directory name too long: \"%s\"", - AH->fSpec); - if (stat(buf, &st) == 0 && S_ISREG(st.st_mode)) - { - AH->format = archDirectory; + AH->format = archDirectory; + if (_fileExistsInDirectory(AH->fSpec, "toc.dat")) return AH->format; - } - #ifdef HAVE_LIBZ - if (snprintf(buf, MAXPGPATH, "%s/toc.dat.gz", AH->fSpec) >= MAXPGPATH) - pg_fatal("directory name too long: \"%s\"", - AH->fSpec); - if (stat(buf, &st) == 0 && S_ISREG(st.st_mode)) - { - AH->format = archDirectory; + if (_fileExistsInDirectory(AH->fSpec, "toc.dat.gz")) return AH->format; - } #endif pg_fatal("directory \"%s\" does not appear to be a valid archive (\"toc.dat\" does not exist)", AH->fSpec); @@ -2178,6 +2181,7 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, SetupWorkerPtrType setupWorkerPtr) { ArchiveHandle *AH; + CompressFileHandle *CFH; pg_compress_specification out_compress_spec = {0}; pg_log_debug("allocating AH for %s, format %d", @@ -2233,7 +2237,10 @@ _allocAH(const char *FileSpec, const ArchiveFormat fmt, /* Open stdout with no compression for AH output handle */ out_compress_spec.algorithm = PG_COMPRESSION_NONE; - AH->OF = cfdopen(dup(fileno(stdout)), PG_BINARY_A, out_compress_spec); + CFH = InitCompressFileHandle(out_compress_spec); + if (CFH->open_func(NULL, fileno(stdout), PG_BINARY_A, CFH)) + pg_fatal("could not open stdout for appending: %m"); + AH->OF = CFH; /* * On Windows, we need to use binary mode to read/write non-text files, @@ -3646,12 +3653,7 @@ WriteHead(ArchiveHandle *AH) AH->WriteBytePtr(AH, AH->intSize); AH->WriteBytePtr(AH, AH->offSize); AH->WriteBytePtr(AH, AH->format); - /* - * For now the compression type is implied by the level. This will need - * to change once support for more compression algorithms is added, - * requiring a format bump. - */ - WriteInt(AH, AH->compression_spec.level); + AH->WriteBytePtr(AH, AH->compression_spec.algorithm); crtm = *localtime(&AH->createDate); WriteInt(AH, crtm.tm_sec); WriteInt(AH, crtm.tm_min); @@ -3722,10 +3724,11 @@ ReadHead(ArchiveHandle *AH) pg_fatal("expected format (%d) differs from format found in file (%d)", AH->format, fmt); - /* Guess the compression method based on the level */ - AH->compression_spec.algorithm = PG_COMPRESSION_NONE; - if (AH->version >= K_VERS_1_2) + if (AH->version >= K_VERS_1_15) + AH->compression_spec.algorithm = AH->ReadBytePtr(AH); + else if (AH->version >= K_VERS_1_2) { + /* Guess the compression method based on the level */ if (AH->version < K_VERS_1_4) AH->compression_spec.level = AH->ReadBytePtr(AH); else @@ -3737,10 +3740,17 @@ ReadHead(ArchiveHandle *AH) else AH->compression_spec.algorithm = PG_COMPRESSION_GZIP; + if (AH->compression_spec.algorithm != PG_COMPRESSION_NONE) + { + bool unsupported = false; + #ifndef HAVE_LIBZ - if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) - pg_fatal("archive is compressed, but this installation does not support compression"); + if (AH->compression_spec.algorithm == PG_COMPRESSION_GZIP) + unsupported = true; #endif + if (unsupported) + pg_fatal("archive is compressed, but this installation does not support compression"); + } if (AH->version >= K_VERS_1_4) { diff --git a/src/bin/pg_dump/pg_backup_archiver.h b/src/bin/pg_dump/pg_backup_archiver.h index 4725e49747..18b38c17ab 100644 --- a/src/bin/pg_dump/pg_backup_archiver.h +++ b/src/bin/pg_dump/pg_backup_archiver.h @@ -65,10 +65,13 @@ #define K_VERS_1_13 MAKE_ARCHIVE_VERSION(1, 13, 0) /* change search_path * behavior */ #define K_VERS_1_14 MAKE_ARCHIVE_VERSION(1, 14, 0) /* add tableam */ +#define K_VERS_1_15 MAKE_ARCHIVE_VERSION(1, 15, 0) /* add + * compression_algorithm + * in header */ /* Current archive version number (the format we can output) */ #define K_VERS_MAJOR 1 -#define K_VERS_MINOR 14 +#define K_VERS_MINOR 15 #define K_VERS_REV 0 #define K_VERS_SELF MAKE_ARCHIVE_VERSION(K_VERS_MAJOR, K_VERS_MINOR, K_VERS_REV) diff --git a/src/bin/pg_dump/pg_backup_custom.c b/src/bin/pg_dump/pg_backup_custom.c index d1e54644a9..512ab043af 100644 --- a/src/bin/pg_dump/pg_backup_custom.c +++ b/src/bin/pg_dump/pg_backup_custom.c @@ -298,7 +298,9 @@ _StartData(ArchiveHandle *AH, TocEntry *te) _WriteByte(AH, BLK_DATA); /* Block type */ WriteInt(AH, te->dumpId); /* For sanity check */ - ctx->cs = AllocateCompressor(AH->compression_spec, _CustomWriteFunc); + ctx->cs = AllocateCompressor(AH->compression_spec, + NULL, + _CustomWriteFunc); } /* @@ -317,15 +319,15 @@ _WriteData(ArchiveHandle *AH, const void *data, size_t dLen) CompressorState *cs = ctx->cs; if (dLen > 0) - /* WriteDataToArchive() internally throws write errors */ - WriteDataToArchive(AH, cs, data, dLen); + /* writeData() internally throws write errors */ + cs->writeData(AH, cs, data, dLen); } /* * Called by the archiver when a dumper's 'DataDumper' routine has * finished. * - * Optional. + * Mandatory. */ static void _EndData(ArchiveHandle *AH, TocEntry *te) @@ -333,6 +335,8 @@ _EndData(ArchiveHandle *AH, TocEntry *te) lclContext *ctx = (lclContext *) AH->formatData; EndCompressor(AH, ctx->cs); + ctx->cs = NULL; + /* Send the end marker */ WriteInt(AH, 0); } @@ -377,7 +381,9 @@ _StartLO(ArchiveHandle *AH, TocEntry *te, Oid oid) WriteInt(AH, oid); - ctx->cs = AllocateCompressor(AH->compression_spec, _CustomWriteFunc); + ctx->cs = AllocateCompressor(AH->compression_spec, + NULL, + _CustomWriteFunc); } /* @@ -566,7 +572,12 @@ _PrintTocData(ArchiveHandle *AH, TocEntry *te) static void _PrintData(ArchiveHandle *AH) { - ReadDataFromArchive(AH, AH->compression_spec, _CustomReadFunc); + CompressorState *cs; + + cs = AllocateCompressor(AH->compression_spec, + _CustomReadFunc, NULL); + cs->readData(AH, cs); + EndCompressor(AH, cs); } static void diff --git a/src/bin/pg_dump/pg_backup_directory.c b/src/bin/pg_dump/pg_backup_directory.c index 6800c3ccee..e2c6e6ecd0 100644 --- a/src/bin/pg_dump/pg_backup_directory.c +++ b/src/bin/pg_dump/pg_backup_directory.c @@ -50,9 +50,8 @@ typedef struct */ char *directory; - cfp *dataFH; /* currently open data file */ - - cfp *LOsTocFH; /* file handle for blobs.toc */ + CompressFileHandle *dataFH; /* currently open data file */ + CompressFileHandle *LOsTocFH; /* file handle for blobs.toc */ ParallelState *pstate; /* for parallel backup / restore */ } lclContext; @@ -198,11 +197,11 @@ InitArchiveFmt_Directory(ArchiveHandle *AH) else { /* Read Mode */ char fname[MAXPGPATH]; - cfp *tocFH; + CompressFileHandle *tocFH; setFilePath(AH, fname, "toc.dat"); - tocFH = cfopen_read(fname, PG_BINARY_R); + tocFH = InitDiscoverCompressFileHandle(fname, PG_BINARY_R); if (tocFH == NULL) pg_fatal("could not open input file \"%s\": %m", fname); @@ -218,7 +217,7 @@ InitArchiveFmt_Directory(ArchiveHandle *AH) ReadToc(AH); /* Nothing else in the file, so close it again... */ - if (cfclose(tocFH) != 0) + if (DestroyCompressFileHandle(tocFH) != 0) pg_fatal("could not close TOC file: %m"); ctx->dataFH = NULL; } @@ -327,9 +326,9 @@ _StartData(ArchiveHandle *AH, TocEntry *te) setFilePath(AH, fname, tctx->filename); - ctx->dataFH = cfopen_write(fname, PG_BINARY_W, - AH->compression_spec); - if (ctx->dataFH == NULL) + ctx->dataFH = InitCompressFileHandle(AH->compression_spec); + + if (ctx->dataFH->open_write_func(fname, PG_BINARY_W, ctx->dataFH)) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -346,15 +345,16 @@ static void _WriteData(ArchiveHandle *AH, const void *data, size_t dLen) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; errno = 0; - if (dLen > 0 && cfwrite(data, dLen, ctx->dataFH) != dLen) + if (dLen > 0 && CFH->write_func(data, dLen, CFH) != dLen) { /* if write didn't set errno, assume problem is no disk space */ if (errno == 0) errno = ENOSPC; pg_fatal("could not write to output file: %s", - get_cfp_error(ctx->dataFH)); + CFH->get_error_func(CFH)); } } @@ -370,7 +370,7 @@ _EndData(ArchiveHandle *AH, TocEntry *te) lclContext *ctx = (lclContext *) AH->formatData; /* Close the file */ - if (cfclose(ctx->dataFH) != 0) + if (DestroyCompressFileHandle(ctx->dataFH) != 0) pg_fatal("could not close data file: %m"); ctx->dataFH = NULL; @@ -385,26 +385,25 @@ _PrintFileData(ArchiveHandle *AH, char *filename) size_t cnt; char *buf; size_t buflen; - cfp *cfp; + CompressFileHandle *CFH; if (!filename) return; - cfp = cfopen_read(filename, PG_BINARY_R); - - if (!cfp) + CFH = InitDiscoverCompressFileHandle(filename, PG_BINARY_R); + if (!CFH) pg_fatal("could not open input file \"%s\": %m", filename); buf = pg_malloc(ZLIB_OUT_SIZE); buflen = ZLIB_OUT_SIZE; - while ((cnt = cfread(buf, buflen, cfp))) + while ((cnt = CFH->read_func(buf, buflen, CFH))) { ahwrite(buf, 1, cnt, AH); } free(buf); - if (cfclose(cfp) != 0) + if (DestroyCompressFileHandle(CFH) != 0) pg_fatal("could not close data file \"%s\": %m", filename); } @@ -435,6 +434,7 @@ _LoadLOs(ArchiveHandle *AH) { Oid oid; lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH; char tocfname[MAXPGPATH]; char line[MAXPGPATH]; @@ -442,14 +442,14 @@ _LoadLOs(ArchiveHandle *AH) setFilePath(AH, tocfname, "blobs.toc"); - ctx->LOsTocFH = cfopen_read(tocfname, PG_BINARY_R); + CFH = ctx->LOsTocFH = InitDiscoverCompressFileHandle(tocfname, PG_BINARY_R); if (ctx->LOsTocFH == NULL) pg_fatal("could not open large object TOC file \"%s\" for input: %m", tocfname); /* Read the LOs TOC file line-by-line, and process each LO */ - while ((cfgets(ctx->LOsTocFH, line, MAXPGPATH)) != NULL) + while ((CFH->gets_func(line, MAXPGPATH, CFH)) != NULL) { char lofname[MAXPGPATH + 1]; char path[MAXPGPATH]; @@ -464,11 +464,11 @@ _LoadLOs(ArchiveHandle *AH) _PrintFileData(AH, path); EndRestoreLO(AH, oid); } - if (!cfeof(ctx->LOsTocFH)) + if (!CFH->eof_func(CFH)) pg_fatal("error reading large object TOC file \"%s\"", tocfname); - if (cfclose(ctx->LOsTocFH) != 0) + if (DestroyCompressFileHandle(ctx->LOsTocFH) != 0) pg_fatal("could not close large object TOC file \"%s\": %m", tocfname); @@ -488,15 +488,16 @@ _WriteByte(ArchiveHandle *AH, const int i) { unsigned char c = (unsigned char) i; lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; errno = 0; - if (cfwrite(&c, 1, ctx->dataFH) != 1) + if (CFH->write_func(&c, 1, CFH) != 1) { /* if write didn't set errno, assume problem is no disk space */ if (errno == 0) errno = ENOSPC; pg_fatal("could not write to output file: %s", - get_cfp_error(ctx->dataFH)); + CFH->get_error_func(CFH)); } return 1; @@ -512,8 +513,9 @@ static int _ReadByte(ArchiveHandle *AH) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; - return cfgetc(ctx->dataFH); + return CFH->getc_func(CFH); } /* @@ -524,15 +526,16 @@ static void _WriteBuf(ArchiveHandle *AH, const void *buf, size_t len) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; errno = 0; - if (cfwrite(buf, len, ctx->dataFH) != len) + if (CFH->write_func(buf, len, CFH) != len) { /* if write didn't set errno, assume problem is no disk space */ if (errno == 0) errno = ENOSPC; pg_fatal("could not write to output file: %s", - get_cfp_error(ctx->dataFH)); + CFH->get_error_func(CFH)); } } @@ -545,12 +548,13 @@ static void _ReadBuf(ArchiveHandle *AH, void *buf, size_t len) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->dataFH; /* - * If there was an I/O error, we already exited in cfread(), so here we + * If there was an I/O error, we already exited in readF(), so here we * exit on short reads. */ - if (cfread(buf, len, ctx->dataFH) != len) + if (CFH->read_func(buf, len, CFH) != len) pg_fatal("could not read from input file: end of file"); } @@ -573,7 +577,7 @@ _CloseArchive(ArchiveHandle *AH) if (AH->mode == archModeWrite) { - cfp *tocFH; + CompressFileHandle *tocFH; pg_compress_specification compression_spec = {0}; char fname[MAXPGPATH]; @@ -584,8 +588,8 @@ _CloseArchive(ArchiveHandle *AH) /* The TOC is always created uncompressed */ compression_spec.algorithm = PG_COMPRESSION_NONE; - tocFH = cfopen_write(fname, PG_BINARY_W, compression_spec); - if (tocFH == NULL) + tocFH = InitCompressFileHandle(compression_spec); + if (tocFH->open_write_func(fname, PG_BINARY_W, tocFH)) pg_fatal("could not open output file \"%s\": %m", fname); ctx->dataFH = tocFH; @@ -598,7 +602,7 @@ _CloseArchive(ArchiveHandle *AH) WriteHead(AH); AH->format = archDirectory; WriteToc(AH); - if (cfclose(tocFH) != 0) + if (DestroyCompressFileHandle(tocFH) != 0) pg_fatal("could not close TOC file: %m"); WriteDataChunks(AH, ctx->pstate); @@ -649,8 +653,8 @@ _StartLOs(ArchiveHandle *AH, TocEntry *te) /* The LO TOC file is never compressed */ compression_spec.algorithm = PG_COMPRESSION_NONE; - ctx->LOsTocFH = cfopen_write(fname, "ab", compression_spec); - if (ctx->LOsTocFH == NULL) + ctx->LOsTocFH = InitCompressFileHandle(compression_spec); + if (ctx->LOsTocFH->open_write_func(fname, "ab", ctx->LOsTocFH)) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -667,9 +671,8 @@ _StartLO(ArchiveHandle *AH, TocEntry *te, Oid oid) snprintf(fname, MAXPGPATH, "%s/blob_%u.dat", ctx->directory, oid); - ctx->dataFH = cfopen_write(fname, PG_BINARY_W, AH->compression_spec); - - if (ctx->dataFH == NULL) + ctx->dataFH = InitCompressFileHandle(AH->compression_spec); + if (ctx->dataFH->open_write_func(fname, PG_BINARY_W, ctx->dataFH)) pg_fatal("could not open output file \"%s\": %m", fname); } @@ -682,18 +685,19 @@ static void _EndLO(ArchiveHandle *AH, TocEntry *te, Oid oid) { lclContext *ctx = (lclContext *) AH->formatData; + CompressFileHandle *CFH = ctx->LOsTocFH; char buf[50]; int len; - /* Close the LO data file itself */ - if (cfclose(ctx->dataFH) != 0) - pg_fatal("could not close LO data file: %m"); + /* Close the BLOB data file itself */ + if (DestroyCompressFileHandle(ctx->dataFH) != 0) + pg_fatal("could not close blob data file: %m"); ctx->dataFH = NULL; /* register the LO in blobs.toc */ len = snprintf(buf, sizeof(buf), "%u blob_%u.dat\n", oid, oid); - if (cfwrite(buf, len, ctx->LOsTocFH) != len) - pg_fatal("could not write to LOs TOC file"); + if (CFH->write_func(buf, len, CFH) != len) + pg_fatal("could not write to blobs TOC file"); } /* @@ -706,8 +710,8 @@ _EndLOs(ArchiveHandle *AH, TocEntry *te) { lclContext *ctx = (lclContext *) AH->formatData; - if (cfclose(ctx->LOsTocFH) != 0) - pg_fatal("could not close LOs TOC file: %m"); + if (DestroyCompressFileHandle(ctx->LOsTocFH) != 0) + pg_fatal("could not close blobs TOC file: %m"); ctx->LOsTocFH = NULL; } diff --git a/src/bin/pg_dump/t/002_pg_dump.pl b/src/bin/pg_dump/t/002_pg_dump.pl index 2eeef2a478..f3ba926321 100644 --- a/src/bin/pg_dump/t/002_pg_dump.pl +++ b/src/bin/pg_dump/t/002_pg_dump.pl @@ -94,7 +94,7 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/compression_gzip_custom.dump", ], - expected => qr/Compression: 1/, + expected => qr/Compression: gzip/, name => 'data content is gzip-compressed' }, }, @@ -239,8 +239,8 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/defaults_custom_format.dump", ], expected => $supports_gzip ? - qr/Compression: -1/ : - qr/Compression: 0/, + qr/Compression: gzip/ : + qr/Compression: none/, name => 'data content is gzip-compressed by default if available', }, }, @@ -264,8 +264,8 @@ my %pgdump_runs = ( command => [ 'pg_restore', '-l', "$tempdir/defaults_dir_format", ], expected => $supports_gzip ? - qr/Compression: -1/ : - qr/Compression: 0/, + qr/Compression: gzip/ : + qr/Compression: none/, name => 'data content is gzip-compressed by default', }, glob_patterns => [ diff --git a/src/tools/pginclude/cpluspluscheck b/src/tools/pginclude/cpluspluscheck index e52fe9f509..da8e4861f4 100755 --- a/src/tools/pginclude/cpluspluscheck +++ b/src/tools/pginclude/cpluspluscheck @@ -150,6 +150,7 @@ do # pg_dump is not C++-clean because it uses "public" and "namespace" # as field names, which is unfortunate but we won't change it now. + test "$f" = src/bin/pg_dump/compress_gzip.h && continue test "$f" = src/bin/pg_dump/compress_io.h && continue test "$f" = src/bin/pg_dump/parallel.h && continue test "$f" = src/bin/pg_dump/pg_backup_archiver.h && continue diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 23bafec5f7..840191d680 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -428,6 +428,7 @@ CompiledExprState CompositeIOData CompositeTypeStmt CompoundAffixFlag +CompressFileHandle CompressionLocation CompressorState ComputeXidHorizonsResult @@ -1034,6 +1035,7 @@ GucStack GucStackState GucStringAssignHook GucStringCheckHook +GzipCompressorState HANDLE HASHACTION HASHBUCKET -- 2.34.1 ^ permalink raw reply [nested|flat] 29+ messages in thread
end of thread, other threads:[~2023-01-16 15:14 UTC | newest] Thread overview: 29+ messages (download: mbox mbox.gz follow: Atom feed) -- links below jump to the message on this page -- 2020-04-08 11:33 [PATCH v9 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v11 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v12 2/7] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2020-07-15 22:35 [PATCH v13 1/6] snapshot scalability: Don't compute global horizons when building snapshots. Andres Freund <[email protected]> 2023-01-14 21:43 Re: Add LZ4 compression in pg_dump Justin Pryzby <[email protected]> 2023-01-15 22:32 ` Re: Add LZ4 compression in pg_dump Justin Pryzby <[email protected]> 2023-01-16 01:28 ` Re: Add LZ4 compression in pg_dump Michael Paquier <[email protected]> 2023-01-16 01:56 ` Re: Add LZ4 compression in pg_dump Justin Pryzby <[email protected]> 2023-01-16 02:54 ` Re: Add LZ4 compression in pg_dump Michael Paquier <[email protected]> 2023-01-16 15:14 ` Re: Add LZ4 compression in pg_dump [email protected]
This inbox is served by agora; see mirroring instructions for how to clone and mirror all data and code used for this inbox