public inbox for [email protected]
help / color / mirror / Atom feed[PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
23+ messages / 2 participants
[nested] [flat]
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension
@ 2020-12-31 12:20 Luc Vlaming <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Luc Vlaming @ 2020-12-31 12:20 UTC (permalink / raw)
---
src/backend/storage/buffer/bufmgr.c | 280 +++++++++++++++++++++++++++-
src/include/storage/buf_internals.h | 2 +-
2 files changed, 275 insertions(+), 7 deletions(-)
diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index d1ce88fee5..14f0923bb5 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -474,6 +474,10 @@ static BufferDesc *BufferAlloc(SMgrRelation smgr,
BlockNumber blockNum,
BufferAccessStrategy strategy,
bool *foundPtr);
+static BufferDesc *BufferAllocExtend(SMgrRelation smgr,
+ char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock);
static void FlushBuffer(BufferDesc *buf, SMgrRelation reln);
static void AtProcExit_Buffers(int code, Datum arg);
static void CheckForBufferLeaks(void);
@@ -996,7 +1000,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
BlockNumber firstBlock,
blockNum;
Block bufBlock;
- bool found;
+ LWLock* lastPartitionLock = NULL;
/* Open it at the smgr level if not already done */
RelationOpenSmgr(reln);
@@ -1007,7 +1011,7 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
LockRelationForExtension(reln, ExclusiveLock);
firstBlock = smgrnblocks(smgr, MAIN_FORKNUM);
- smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, false, BULK_INSERT_BATCH_SIZE);
+ smgrextend_count(smgr, MAIN_FORKNUM, firstBlock, bistate->empty_buffer, true, BULK_INSERT_BATCH_SIZE);
UnlockRelationForExtension(reln, ExclusiveLock);
for (int i=0; i<BULK_INSERT_BATCH_SIZE; ++i)
@@ -1017,12 +1021,10 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
pgstat_count_buffer_read(reln);
/* Make sure we will have room to remember the buffer pin */
ResourceOwnerEnlargeBuffers(CurrentResourceOwner);
- bufHdr = BufferAlloc(smgr, relpersistence, MAIN_FORKNUM, blockNum,
- bistate->strategy, &found);
+ bufHdr = BufferAllocExtend(smgr, relpersistence, MAIN_FORKNUM, blockNum,
+ bistate->strategy, &lastPartitionLock);
pgBufferUsage.shared_blks_written++;
- Assert(!found);
-
bufBlock = BufHdrGetBlock(bufHdr);
/* new buffers are zero-filled */
@@ -1033,6 +1035,9 @@ ReadBufferExtendBulk(Relation reln, struct BulkInsertStateData* bistate)
bistate->local_buffers[i] = BufferDescriptorGetBuffer(bufHdr);
}
+ Assert(lastPartitionLock);
+ LWLockRelease(lastPartitionLock);
+
bistate->local_buffers_idx = 0;
VacuumPageMiss += BULK_INSERT_BATCH_SIZE;
@@ -1408,6 +1413,269 @@ BufferAlloc(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
return buf;
}
+static BufferDesc *
+BufferAllocExtend(SMgrRelation smgr, char relpersistence, ForkNumber forkNum,
+ BlockNumber blockNum, BufferAccessStrategy strategy,
+ LWLock** lastPartitionLock)
+{
+ BufferTag newTag; /* identity of requested block */
+ uint32 newHash; /* hash value for newTag */
+ LWLock *newPartitionLock; /* buffer partition lock for it */
+ BufferTag oldTag; /* previous identity of selected buffer */
+ uint32 oldHash; /* hash value for oldTag */
+ LWLock *oldPartitionLock; /* buffer partition lock for it */
+ uint32 oldFlags;
+ int buf_id;
+ BufferDesc *buf;
+ uint32 buf_state;
+
+ Assert(lastPartitionLock);
+
+ /* create a tag so we can lookup the buffer */
+ INIT_BUFFERTAG(newTag, smgr->smgr_rnode.node, forkNum, blockNum);
+
+ /* determine its hash code and partition lock ID */
+ newHash = BufTableHashCode(&newTag);
+ newPartitionLock = BufMappingPartitionLock(newHash);
+
+ /* Loop here in case we have to try another victim buffer */
+ for (;;)
+ {
+ /*
+ * Ensure, while the spinlock's not yet held, that there's a free
+ * refcount entry.
+ */
+ ReservePrivateRefCountEntry();
+
+ /*
+ * Select a victim buffer. The buffer is returned with its header
+ * spinlock still held!
+ */
+ buf = StrategyGetBuffer(strategy, &buf_state);
+
+ Assert(BUF_STATE_GET_REFCOUNT(buf_state) == 0);
+
+ /* Must copy buffer flags while we still hold the spinlock */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+
+ /* Pin the buffer and then release the buffer spinlock */
+ PinBuffer_Locked(buf);
+
+ /*
+ * If the buffer was dirty, try to write it out. There is a race
+ * condition here, in that someone might dirty it after we released it
+ * above, or even while we are writing it out (since our share-lock
+ * won't prevent hint-bit updates). We will recheck the dirty bit
+ * after re-locking the buffer header.
+ */
+ if (oldFlags & BM_DIRTY)
+ {
+ /*
+ * We need a share-lock on the buffer contents to write it out
+ * (else we might write invalid data, eg because someone else is
+ * compacting the page contents while we write). We must use a
+ * conditional lock acquisition here to avoid deadlock. Even
+ * though the buffer was not pinned (and therefore surely not
+ * locked) when StrategyGetBuffer returned it, someone else could
+ * have pinned and exclusive-locked it by the time we get here. If
+ * we try to get the lock unconditionally, we'd block waiting for
+ * them; if they later block waiting for us, deadlock ensues.
+ * (This has been observed to happen when two backends are both
+ * trying to split btree index pages, and the second one just
+ * happens to be trying to split the page the first one got from
+ * StrategyGetBuffer.)
+ */
+ if (LWLockConditionalAcquire(BufferDescriptorGetContentLock(buf),
+ LW_SHARED))
+ {
+ /*
+ * If using a nondefault strategy, and writing the buffer
+ * would require a WAL flush, let the strategy decide whether
+ * to go ahead and write/reuse the buffer or to choose another
+ * victim. We need lock to inspect the page LSN, so this
+ * can't be done inside StrategyGetBuffer.
+ */
+ if (strategy != NULL)
+ {
+ XLogRecPtr lsn;
+
+ /* Read the LSN while holding buffer header lock */
+ buf_state = LockBufHdr(buf);
+ lsn = BufferGetLSN(buf);
+ UnlockBufHdr(buf, buf_state);
+
+ if (XLogNeedsFlush(lsn) &&
+ StrategyRejectBuffer(strategy, buf))
+ {
+ /* Drop lock/pin and loop around for another buffer */
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /* OK, do the I/O */
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_START(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+
+ FlushBuffer(buf, NULL);
+ LWLockRelease(BufferDescriptorGetContentLock(buf));
+
+ ScheduleBufferTagForWriteback(&BackendWritebackContext,
+ &buf->tag);
+
+ TRACE_POSTGRESQL_BUFFER_WRITE_DIRTY_DONE(forkNum, blockNum,
+ smgr->smgr_rnode.node.spcNode,
+ smgr->smgr_rnode.node.dbNode,
+ smgr->smgr_rnode.node.relNode);
+ }
+ else
+ {
+ /*
+ * Someone else has locked the buffer, so give it up and loop
+ * back to get another one.
+ */
+ UnpinBuffer(buf, true);
+ continue;
+ }
+ }
+
+ /*
+ * To change the association of a valid buffer, we'll need to have
+ * exclusive lock on both the old and new mapping partitions.
+ */
+ if (oldFlags & BM_TAG_VALID)
+ {
+ /*
+ * Need to compute the old tag's hashcode and partition lock ID.
+ * XXX is it worth storing the hashcode in BufferDesc so we need
+ * not recompute it here? Probably not.
+ */
+ oldTag = buf->tag;
+ oldHash = BufTableHashCode(&oldTag);
+ oldPartitionLock = BufMappingPartitionLock(oldHash);
+
+ if (*lastPartitionLock &&
+ (*lastPartitionLock != oldPartitionLock ||
+ *lastPartitionLock != newPartitionLock))
+ {
+ LWLockRelease(*lastPartitionLock);
+ *lastPartitionLock = NULL;
+ }
+
+ /*
+ * Must lock the lower-numbered partition first to avoid
+ * deadlocks.
+ */
+ if (oldPartitionLock < newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ else if (oldPartitionLock > newPartitionLock)
+ {
+ Assert(*lastPartitionLock == NULL);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ LWLockAcquire(oldPartitionLock, LW_EXCLUSIVE);
+ }
+ else
+ {
+ /* only one partition, only one lock */
+ if (*lastPartitionLock != newPartitionLock)
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+ }
+ else
+ {
+ /* if it wasn't valid, we need only the new partition */
+ if (*lastPartitionLock != newPartitionLock)
+ {
+ if (*lastPartitionLock)
+ LWLockRelease(*lastPartitionLock);
+ LWLockAcquire(newPartitionLock, LW_EXCLUSIVE);
+ }
+
+ /* remember we have no old-partition lock or tag */
+ oldPartitionLock = NULL;
+ /* keep the compiler quiet about uninitialized variables */
+ oldHash = 0;
+ }
+
+ *lastPartitionLock = newPartitionLock;
+
+ /*
+ * Try to make a hashtable entry for the buffer under its new tag.
+ * This could fail because while we were writing someone else
+ * allocated another buffer for the same block we want to read in.
+ * Note that we have not yet removed the hashtable entry for the old
+ * tag.
+ */
+ buf_id = BufTableInsert(&newTag, newHash, buf->buf_id);
+ Assert(buf_id < 0);
+
+ /*
+ * Need to lock the buffer header too in order to change its tag.
+ */
+ buf_state = LockBufHdr(buf);
+
+ /*
+ * Somebody could have pinned or re-dirtied the buffer while we were
+ * doing the I/O and making the new hashtable entry. If so, we can't
+ * recycle this buffer; we must undo everything we've done and start
+ * over with a new victim buffer.
+ */
+ oldFlags = buf_state & BUF_FLAG_MASK;
+ if (BUF_STATE_GET_REFCOUNT(buf_state) == 1 && !(oldFlags & BM_DIRTY))
+ break;
+
+ pg_unreachable();
+ }
+
+ /*
+ * Okay, it's finally safe to rename the buffer.
+ *
+ * Clearing BM_VALID here is necessary, clearing the dirtybits is just
+ * paranoia. We also reset the usage_count since any recency of use of
+ * the old content is no longer relevant. (The usage_count starts out at
+ * 1 so that the buffer can survive one clock-sweep pass.)
+ *
+ * Make sure BM_PERMANENT is set for buffers that must be written at every
+ * checkpoint. Unlogged buffers only need to be written at shutdown
+ * checkpoints, except for their "init" forks, which need to be treated
+ * just like permanent relations.
+ */
+ buf->tag = newTag;
+ buf_state &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED |
+ BM_CHECKPOINT_NEEDED | BM_IO_ERROR | BM_PERMANENT |
+ BUF_USAGECOUNT_MASK);
+ if (relpersistence == RELPERSISTENCE_PERMANENT || forkNum == INIT_FORKNUM)
+ buf_state |= BM_TAG_VALID | BM_PERMANENT | BUF_USAGECOUNT_ONE;
+ else
+ buf_state |= BM_TAG_VALID | BUF_USAGECOUNT_ONE;
+
+ UnlockBufHdr(buf, buf_state);
+
+ if (oldPartitionLock != NULL)
+ {
+ BufTableDelete(&oldTag, oldHash);
+ if (oldPartitionLock != newPartitionLock)
+ LWLockRelease(oldPartitionLock);
+ }
+
+ /*
+ * Buffer contents are currently invalid. Try to get the io_in_progress
+ * lock. If StartBufferIO returns false, then someone else managed to
+ * read it before we did, so there's nothing left for BufferAlloc() to do.
+ */
+ StartBufferIO(buf, true);
+
+ return buf;
+}
+
+
/*
* InvalidateBuffer -- mark a shared buffer invalid and return it to the
* freelist.
diff --git a/src/include/storage/buf_internals.h b/src/include/storage/buf_internals.h
index 3377fa5676..277705f18c 100644
--- a/src/include/storage/buf_internals.h
+++ b/src/include/storage/buf_internals.h
@@ -124,7 +124,7 @@ typedef struct buftag
* NB: NUM_BUFFER_PARTITIONS must be a power of 2!
*/
#define BufTableHashPartition(hashcode) \
- ((hashcode) % NUM_BUFFER_PARTITIONS)
+ ((hashcode >> 7) % NUM_BUFFER_PARTITIONS)
#define BufMappingPartitionLock(hashcode) \
(&MainLWLockArray[BUFFER_MAPPING_LWLOCK_OFFSET + \
BufTableHashPartition(hashcode)].lock)
--
2.25.1
--------------065E078E0383D7C686F94091--
^ permalink raw reply [nested|flat] 23+ messages in thread
* Re: Strange presentaion related to inheritance in \d+
@ 2023-08-28 11:36 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 23+ messages in thread
From: Alvaro Herrera @ 2023-08-28 11:36 UTC (permalink / raw)
To: Kyotaro Horiguchi <[email protected]>; +Cc: [email protected]
On 2023-Aug-28, Kyotaro Horiguchi wrote:
> But with these tables:
>
> create table p (a int, b int not null default 0);
> create table c1 (a int, b int not null NO INHERIT default 1) inherits (p);
>
> I get:
>
> > Not-null constraints:
> > "c1_b_not_null" NOT NULL "b" *NO INHERIT*
>
> Here, "NO INHERIT" is mapped from connoinherit, and conislocal and
> "coninhcount <> 0" align with "local" and "inherited". For a clearer
> picuture, those values for c1 are as follows.
Hmm, I think the bug here is that we let you create a constraint in c1
that is NOINHERIT. If the parent already has one INHERIT constraint
in that column, then the child must have that one also; it's not
possible to have both a constraint that inherits and one that doesn't.
I understand that there are only three possibilities for a NOT NULL
constraint in a column:
- There's a NO INHERIT constraint. A NO INHERIT constraint is always
defined locally in that table. In this case, if there is a parent
relation, then it must either not have a NOT NULL constraint in that
column, or it may also have a NO INHERIT one. Therefore, it's
correct to print NO INHERIT and nothing else. We could also print
"(local)" but I see no point in doing that.
- A constraint comes inherited from one or more parent tables and has no
local definition. In this case, the constraint always inherits
(otherwise, the parent wouldn't have given it to this table). So
printing "(inherited)" and nothing else is correct.
- A constraint can have a local definition and also be inherited. In
this case, printing "(local, inherited)" is correct.
Have I missed other cases?
The NO INHERIT bit is part of the syntax, which is why I put it in
uppercase and not marked it for translation. The other two are
informational, so they are translatable.
--
Álvaro Herrera PostgreSQL Developer — https://www.EnterpriseDB.com/
"The important things in the world are problems with society that we don't
understand at all. The machines will become more complicated but they won't
be more complicated than the societies that run them." (Freeman Dyson)
^ permalink raw reply [nested|flat] 23+ messages in thread
end of thread, other threads:[~2023-08-28 11:36 UTC | newest]
Thread overview: 23+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2020-12-31 12:20 [PATCH v1 2/2] WIP: buffer alloc specialized for relation extension Luc Vlaming <[email protected]>
2023-08-28 11:36 Re: Strange presentaion related to inheritance in \d+ Alvaro Herrera <[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