agora inbox for [email protected]  

help / color / mirror / Atom feed

[PATCH] Unify error messages
1055+ messages / 4 participants
[nested] [flat]

* [PATCH] Unify error messages
@ 2019-04-22 19:45  Alvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Alvaro Herrera @ 2019-04-22 19:45 UTC (permalink / raw)

... for translatability purposes.
---
 src/backend/storage/ipc/latch.c       | 12 +++++++++---
 src/backend/storage/ipc/signalfuncs.c |  4 +++-
 src/backend/utils/adt/formatting.c    |  9 ++++++---
 src/backend/utils/adt/genfile.c       |  4 +++-
 src/backend/utils/adt/json.c          |  4 +++-
 src/backend/utils/adt/jsonb.c         |  4 +++-
 6 files changed, 27 insertions(+), 10 deletions(-)

diff --git a/src/backend/storage/ipc/latch.c b/src/backend/storage/ipc/latch.c
index 59fa917ae04..e0712f906a1 100644
--- a/src/backend/storage/ipc/latch.c
+++ b/src/backend/storage/ipc/latch.c
@@ -856,7 +856,9 @@ WaitEventAdjustEpoll(WaitEventSet *set, WaitEvent *event, int action)
 	if (rc < 0)
 		ereport(ERROR,
 				(errcode_for_socket_access(),
-				 errmsg("epoll_ctl() failed: %m")));
+				 /* translator: %s is a syscall name, such as "poll()" */
+				 errmsg("%s failed: %m",
+						"epoll_ctl()")));
 }
 #endif
 
@@ -1087,7 +1089,9 @@ WaitEventSetWaitBlock(WaitEventSet *set, int cur_timeout,
 			waiting = false;
 			ereport(ERROR,
 					(errcode_for_socket_access(),
-					 errmsg("epoll_wait() failed: %m")));
+					 /* translator: %s is a syscall name, such as "poll()" */
+					 errmsg("%s failed: %m",
+							"epoll_wait()")));
 		}
 		return 0;
 	}
@@ -1211,7 +1215,9 @@ WaitEventSetWaitBlock(WaitEventSet *set, int cur_timeout,
 			waiting = false;
 			ereport(ERROR,
 					(errcode_for_socket_access(),
-					 errmsg("poll() failed: %m")));
+					 /* translator: %s is a syscall name, such as "poll()" */
+					 errmsg("%s failed: %m",
+							"poll()")));
 		}
 		return 0;
 	}
diff --git a/src/backend/storage/ipc/signalfuncs.c b/src/backend/storage/ipc/signalfuncs.c
index 4769b1b51eb..4bfbd57464c 100644
--- a/src/backend/storage/ipc/signalfuncs.c
+++ b/src/backend/storage/ipc/signalfuncs.c
@@ -181,7 +181,9 @@ pg_rotate_logfile(PG_FUNCTION_ARGS)
 		ereport(ERROR,
 				(errcode(ERRCODE_INSUFFICIENT_PRIVILEGE),
 				 (errmsg("must be superuser to rotate log files with adminpack 1.0"),
-				  errhint("Consider using pg_logfile_rotate(), which is part of core, instead."))));
+				  /* translator: %s is a SQL function name */
+				  errhint("Consider using %s, which is part of core, instead.",
+						  "pg_logfile_rotate()"))));
 
 	if (!Logging_collector)
 	{
diff --git a/src/backend/utils/adt/formatting.c b/src/backend/utils/adt/formatting.c
index df1db7bc9f1..69a691f18e7 100644
--- a/src/backend/utils/adt/formatting.c
+++ b/src/backend/utils/adt/formatting.c
@@ -1566,7 +1566,8 @@ str_tolower(const char *buff, size_t nbytes, Oid collid)
 				 */
 				ereport(ERROR,
 						(errcode(ERRCODE_INDETERMINATE_COLLATION),
-						 errmsg("could not determine which collation to use for lower() function"),
+						 errmsg("could not determine which collation to use for %s function",
+								"lower()"),
 						 errhint("Use the COLLATE clause to set the collation explicitly.")));
 			}
 			mylocale = pg_newlocale_from_collation(collid);
@@ -1688,7 +1689,8 @@ str_toupper(const char *buff, size_t nbytes, Oid collid)
 				 */
 				ereport(ERROR,
 						(errcode(ERRCODE_INDETERMINATE_COLLATION),
-						 errmsg("could not determine which collation to use for upper() function"),
+						 errmsg("could not determine which collation to use for %s function",
+								"upper()"),
 						 errhint("Use the COLLATE clause to set the collation explicitly.")));
 			}
 			mylocale = pg_newlocale_from_collation(collid);
@@ -1811,7 +1813,8 @@ str_initcap(const char *buff, size_t nbytes, Oid collid)
 				 */
 				ereport(ERROR,
 						(errcode(ERRCODE_INDETERMINATE_COLLATION),
-						 errmsg("could not determine which collation to use for initcap() function"),
+						 errmsg("could not determine which collation to use for %s function",
+								"initcap()"),
 						 errhint("Use the COLLATE clause to set the collation explicitly.")));
 			}
 			mylocale = pg_newlocale_from_collation(collid);
diff --git a/src/backend/utils/adt/genfile.c b/src/backend/utils/adt/genfile.c
index d6976609968..a3c6adaf640 100644
--- a/src/backend/utils/adt/genfile.c
+++ b/src/backend/utils/adt/genfile.c
@@ -219,7 +219,9 @@ pg_read_file(PG_FUNCTION_ARGS)
 		ereport(ERROR,
 				(errcode(ERRCODE_INSUFFICIENT_PRIVILEGE),
 				 (errmsg("must be superuser to read files with adminpack 1.0"),
-				  errhint("Consider using pg_file_read(), which is part of core, instead."))));
+				  /* translator: %s is a SQL function name */
+				  errhint("Consider using %s, which is part of core, instead.",
+						  "pg_file_read()"))));
 
 	/* handle optional arguments */
 	if (PG_NARGS() >= 3)
diff --git a/src/backend/utils/adt/json.c b/src/backend/utils/adt/json.c
index de0d0723b71..bb4bac85f7d 100644
--- a/src/backend/utils/adt/json.c
+++ b/src/backend/utils/adt/json.c
@@ -2192,7 +2192,9 @@ json_build_object(PG_FUNCTION_ARGS)
 		ereport(ERROR,
 				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
 				 errmsg("argument list must have even number of elements"),
-				 errhint("The arguments of json_build_object() must consist of alternating keys and values.")));
+				 /* translator: %s is a SQL function name */
+				 errhint("The arguments of %s must consist of alternating keys and values.",
+						 "json_build_object()")));
 
 	result = makeStringInfo();
 
diff --git a/src/backend/utils/adt/jsonb.c b/src/backend/utils/adt/jsonb.c
index 7af4091200b..036d771386f 100644
--- a/src/backend/utils/adt/jsonb.c
+++ b/src/backend/utils/adt/jsonb.c
@@ -1155,7 +1155,9 @@ jsonb_build_object(PG_FUNCTION_ARGS)
 		ereport(ERROR,
 				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
 				 errmsg("argument list must have even number of elements"),
-				 errhint("The arguments of jsonb_build_object() must consist of alternating keys and values.")));
+				 /* translator: %s is a SQL function name */
+				 errhint("The arguments of %s must consist of alternating keys and values.",
+						 "jsonb_build_object()")));
 
 	memset(&result, 0, sizeof(JsonbInState));
 
-- 
2.17.1


--W/nzBZO5zC0uMSeA--





^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v1 1/2] partial revert of ff9618e82a
@ 2023-06-14 17:54  Nathan Bossart <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Nathan Bossart @ 2023-06-14 17:54 UTC (permalink / raw)

---
 doc/src/sgml/ref/analyze.sgml                 |  5 +---
 doc/src/sgml/ref/cluster.sgml                 |  6 +---
 doc/src/sgml/ref/lock.sgml                    |  5 +---
 doc/src/sgml/ref/reindex.sgml                 |  6 +---
 doc/src/sgml/ref/vacuum.sgml                  |  5 +---
 src/backend/commands/cluster.c                | 10 ++-----
 src/backend/commands/indexcmds.c              | 19 +++++-------
 src/backend/commands/lockcmds.c               |  8 -----
 src/backend/commands/tablecmds.c              | 29 +------------------
 src/backend/commands/vacuum.c                 |  6 +---
 src/include/commands/tablecmds.h              |  1 -
 .../expected/cluster-conflict-partition.out   | 14 ++++-----
 .../specs/cluster-conflict-partition.spec     |  7 +++--
 src/test/regress/expected/cluster.out         |  3 +-
 src/test/regress/expected/vacuum.out          | 18 ++++++++++++
 15 files changed, 49 insertions(+), 93 deletions(-)

diff --git a/doc/src/sgml/ref/analyze.sgml b/doc/src/sgml/ref/analyze.sgml
index 20c6f9939f..30a893230e 100644
--- a/doc/src/sgml/ref/analyze.sgml
+++ b/doc/src/sgml/ref/analyze.sgml
@@ -190,10 +190,7 @@ ANALYZE [ VERBOSE ] [ <replaceable class="parameter">table_and_columns</replacea
    analyze all tables in their databases, except shared catalogs.
    (The restriction for shared catalogs means that a true database-wide
    <command>ANALYZE</command> can only be performed by superusers and roles
-   with privileges of <literal>pg_maintain</literal>.)  If a role has
-   permission to <command>ANALYZE</command> a partitioned table, it is also
-   permitted to <command>ANALYZE</command> each of its partitions, regardless
-   of whether the role has the aforementioned privileges on the partition.
+   with privileges of <literal>pg_maintain</literal>.)
    <command>ANALYZE</command> will skip over any tables that the calling user
    does not have permission to analyze.
   </para>
diff --git a/doc/src/sgml/ref/cluster.sgml b/doc/src/sgml/ref/cluster.sgml
index 29f0f1fd90..c768175499 100644
--- a/doc/src/sgml/ref/cluster.sgml
+++ b/doc/src/sgml/ref/cluster.sgml
@@ -137,11 +137,7 @@ CLUSTER [VERBOSE]
     on the table or be the table's owner, a superuser, or a role with
     privileges of the
     <link linkend="predefined-roles-table"><literal>pg_maintain</literal></link>
-    role.  If a role has permission to <command>CLUSTER</command> a partitioned
-    table, it is also permitted to <command>CLUSTER</command> each of its
-    partitions, regardless of whether the role has the aforementioned
-    privileges on the partition.  <command>CLUSTER</command> will skip over any
-    tables that the calling user does not have permission to cluster.
+    role.
    </para>
 
    <para>
diff --git a/doc/src/sgml/ref/lock.sgml b/doc/src/sgml/ref/lock.sgml
index 5b3b2b793a..8524182211 100644
--- a/doc/src/sgml/ref/lock.sgml
+++ b/doc/src/sgml/ref/lock.sgml
@@ -177,10 +177,7 @@ LOCK [ TABLE ] [ ONLY ] <replaceable class="parameter">name</replaceable> [ * ]
     MODE</literal> (or a less-conflicting mode as described in <xref
     linkend="explicit-locking"/>) is permitted. If a user has
     <literal>SELECT</literal> privileges on the table, <literal>ACCESS SHARE
-    MODE</literal> is permitted.  If a role has permission to lock a
-    partitioned table, it is also permitted to lock each of its partitions,
-    regardless of whether the role has the aforementioned privileges on the
-    partition.
+    MODE</literal> is permitted.
    </para>
 
    <para>
diff --git a/doc/src/sgml/ref/reindex.sgml b/doc/src/sgml/ref/reindex.sgml
index 71455dfdc7..23f8c7630b 100644
--- a/doc/src/sgml/ref/reindex.sgml
+++ b/doc/src/sgml/ref/reindex.sgml
@@ -306,11 +306,7 @@ REINDEX [ ( <replaceable class="parameter">option</replaceable> [, ...] ) ] { DA
    indexes on shared catalogs will be skipped unless the user owns the
    catalog (which typically won't be the case), has privileges of the
    <literal>pg_maintain</literal> role, or has the <literal>MAINTAIN</literal>
-   privilege on the catalog.  If a role has permission to
-   <command>REINDEX</command> a partitioned table, it is also permitted to
-   <command>REINDEX</command> each of its partitions, regardless of whether the
-   role has the aforementioned privileges on the partition.  Of course,
-   superusers can always reindex anything.
+   privilege on the catalog.  Of course, superusers can always reindex anything.
   </para>
 
   <para>
diff --git a/doc/src/sgml/ref/vacuum.sgml b/doc/src/sgml/ref/vacuum.sgml
index 57bc4c23ec..445325e14c 100644
--- a/doc/src/sgml/ref/vacuum.sgml
+++ b/doc/src/sgml/ref/vacuum.sgml
@@ -452,10 +452,7 @@ VACUUM [ FULL ] [ FREEZE ] [ VERBOSE ] [ ANALYZE ] [ <replaceable class="paramet
     vacuum all tables in their databases, except shared catalogs.
     (The restriction for shared catalogs means that a true database-wide
     <command>VACUUM</command> can only be performed by superusers and roles
-    with privileges of <literal>pg_maintain</literal>.)  If a role has
-    permission to <command>VACUUM</command> a partitioned table, it is also
-    permitted to <command>VACUUM</command> each of its partitions, regardless
-    of whether the role has the aforementioned privileges on the partition.
+    with privileges of <literal>pg_maintain</literal>.)
     <command>VACUUM</command> will skip over any tables that the calling user
     does not have permission to vacuum.
    </para>
diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 369fea7c04..38834356b9 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -1693,11 +1693,8 @@ get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
 		if (get_rel_relkind(indexrelid) != RELKIND_INDEX)
 			continue;
 
-		/*
-		 * We already checked that the user has privileges to CLUSTER the
-		 * partitioned table when we locked it earlier, so there's no need to
-		 * check the privileges again here.
-		 */
+		if (!cluster_is_permitted_for_relation(relid, GetUserId()))
+			continue;
 
 		/* Use a permanent memory context for the result list */
 		old_context = MemoryContextSwitchTo(cluster_context);
@@ -1720,8 +1717,7 @@ get_tables_to_cluster_partitioned(MemoryContext cluster_context, Oid indexOid)
 static bool
 cluster_is_permitted_for_relation(Oid relid, Oid userid)
 {
-	if (pg_class_aclcheck(relid, userid, ACL_MAINTAIN) == ACLCHECK_OK ||
-		has_partition_ancestor_privs(relid, userid, ACL_MAINTAIN))
+	if (pg_class_aclcheck(relid, userid, ACL_MAINTAIN) == ACLCHECK_OK)
 		return true;
 
 	ereport(WARNING,
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index a5168c9f09..56bba91c08 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -2854,8 +2854,7 @@ RangeVarCallbackForReindexIndex(const RangeVar *relation,
 	/* Check permissions */
 	table_oid = IndexGetRelation(relId, true);
 	if (OidIsValid(table_oid) &&
-		pg_class_aclcheck(table_oid, GetUserId(), ACL_MAINTAIN) != ACLCHECK_OK &&
-		!has_partition_ancestor_privs(table_oid, GetUserId(), ACL_MAINTAIN))
+		pg_class_aclcheck(table_oid, GetUserId(), ACL_MAINTAIN) != ACLCHECK_OK)
 		aclcheck_error(ACLCHECK_NOT_OWNER, OBJECT_INDEX,
 					   relation->relname);
 
@@ -3065,17 +3064,15 @@ ReindexMultipleTables(const char *objectName, ReindexObjectType objectKind,
 
 		/*
 		 * The table can be reindexed if the user has been granted MAINTAIN on
-		 * the table or one of its partition ancestors or the user is a
-		 * superuser, the table owner, or the database/schema owner (but in
-		 * the latter case, only if it's not a shared relation).
-		 * pg_class_aclcheck includes the superuser case, and depending on
-		 * objectKind we already know that the user has permission to run
-		 * REINDEX on this database or schema per the permission checks at the
-		 * beginning of this routine.
+		 * the table or the user is a superuser, the table owner, or the
+		 * database/schema owner (but in the latter case, only if it's not a
+		 * shared relation).  pg_class_aclcheck includes the superuser case,
+		 * and depending on objectKind we already know that the user has
+		 * permission to run REINDEX on this database or schema per the
+		 * permission checks at the beginning of this routine.
 		 */
 		if (classtuple->relisshared &&
-			pg_class_aclcheck(relid, GetUserId(), ACL_MAINTAIN) != ACLCHECK_OK &&
-			!has_partition_ancestor_privs(relid, GetUserId(), ACL_MAINTAIN))
+			pg_class_aclcheck(relid, GetUserId(), ACL_MAINTAIN) != ACLCHECK_OK)
 			continue;
 
 		/*
diff --git a/src/backend/commands/lockcmds.c b/src/backend/commands/lockcmds.c
index 43c7d7f4bb..92662cbbc8 100644
--- a/src/backend/commands/lockcmds.c
+++ b/src/backend/commands/lockcmds.c
@@ -19,7 +19,6 @@
 #include "catalog/namespace.h"
 #include "catalog/pg_inherits.h"
 #include "commands/lockcmds.h"
-#include "commands/tablecmds.h"
 #include "miscadmin.h"
 #include "nodes/nodeFuncs.h"
 #include "parser/parse_clause.h"
@@ -297,12 +296,5 @@ LockTableAclCheck(Oid reloid, LOCKMODE lockmode, Oid userid)
 
 	aclresult = pg_class_aclcheck(reloid, userid, aclmask);
 
-	/*
-	 * If this is a partition, check permissions of its ancestors if needed.
-	 */
-	if (aclresult != ACLCHECK_OK &&
-		has_partition_ancestor_privs(reloid, userid, ACL_MAINTAIN))
-		aclresult = ACLCHECK_OK;
-
 	return aclresult;
 }
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 4d49d70c33..29eece1c2c 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -17006,38 +17006,11 @@ RangeVarCallbackMaintainsTable(const RangeVar *relation,
 				 errmsg("\"%s\" is not a table or materialized view", relation->relname)));
 
 	/* Check permissions */
-	if (pg_class_aclcheck(relId, GetUserId(), ACL_MAINTAIN) != ACLCHECK_OK &&
-		!has_partition_ancestor_privs(relId, GetUserId(), ACL_MAINTAIN))
+	if (pg_class_aclcheck(relId, GetUserId(), ACL_MAINTAIN) != ACLCHECK_OK)
 		aclcheck_error(ACLCHECK_NOT_OWNER, OBJECT_TABLE,
 					   relation->relname);
 }
 
-/*
- * If relid is a partition, returns whether userid has any of the privileges
- * specified in acl on any of its ancestors.  Otherwise, returns false.
- */
-bool
-has_partition_ancestor_privs(Oid relid, Oid userid, AclMode acl)
-{
-	List	   *ancestors;
-	ListCell   *lc;
-
-	if (!get_rel_relispartition(relid))
-		return false;
-
-	ancestors = get_partition_ancestors(relid);
-	foreach(lc, ancestors)
-	{
-		Oid			ancestor = lfirst_oid(lc);
-
-		if (OidIsValid(ancestor) &&
-			pg_class_aclcheck(ancestor, userid, acl) == ACLCHECK_OK)
-			return true;
-	}
-
-	return false;
-}
-
 /*
  * Callback to RangeVarGetRelidExtended() for TRUNCATE processing.
  */
diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c
index a843f9ad92..2e41a31173 100644
--- a/src/backend/commands/vacuum.c
+++ b/src/backend/commands/vacuum.c
@@ -41,7 +41,6 @@
 #include "catalog/pg_namespace.h"
 #include "commands/cluster.h"
 #include "commands/defrem.h"
-#include "commands/tablecmds.h"
 #include "commands/vacuum.h"
 #include "miscadmin.h"
 #include "nodes/makefuncs.h"
@@ -719,13 +718,10 @@ vacuum_is_permitted_for_relation(Oid relid, Form_pg_class reltuple,
 	 *   - the role owns the relation
 	 *   - the role owns the current database and the relation is not shared
 	 *   - the role has been granted the MAINTAIN privilege on the relation
-	 *   - the role has privileges to vacuum/analyze any of the relation's
-	 *     partition ancestors
 	 *----------
 	 */
 	if ((object_ownercheck(DatabaseRelationId, MyDatabaseId, GetUserId()) && !reltuple->relisshared) ||
-		pg_class_aclcheck(relid, GetUserId(), ACL_MAINTAIN) == ACLCHECK_OK ||
-		has_partition_ancestor_privs(relid, GetUserId(), ACL_MAINTAIN))
+		pg_class_aclcheck(relid, GetUserId(), ACL_MAINTAIN) == ACLCHECK_OK)
 		return true;
 
 	relname = NameStr(reltuple->relname);
diff --git a/src/include/commands/tablecmds.h b/src/include/commands/tablecmds.h
index 17b9404937..250d89ff88 100644
--- a/src/include/commands/tablecmds.h
+++ b/src/include/commands/tablecmds.h
@@ -99,7 +99,6 @@ extern void AtEOSubXact_on_commit_actions(bool isCommit,
 extern void RangeVarCallbackMaintainsTable(const RangeVar *relation,
 										   Oid relId, Oid oldRelId,
 										   void *arg);
-extern bool has_partition_ancestor_privs(Oid relid, Oid userid, AclMode acl);
 
 extern void RangeVarCallbackOwnsRelation(const RangeVar *relation,
 										 Oid relId, Oid oldRelId, void *arg);
diff --git a/src/test/isolation/expected/cluster-conflict-partition.out b/src/test/isolation/expected/cluster-conflict-partition.out
index 8d21276996..7be9e56ef1 100644
--- a/src/test/isolation/expected/cluster-conflict-partition.out
+++ b/src/test/isolation/expected/cluster-conflict-partition.out
@@ -3,7 +3,7 @@ Parsed test spec with 2 sessions
 starting permutation: s1_begin s1_lock_parent s2_auth s2_cluster s1_commit s2_reset
 step s1_begin: BEGIN;
 step s1_lock_parent: LOCK cluster_part_tab IN SHARE UPDATE EXCLUSIVE MODE;
-step s2_auth: SET ROLE regress_cluster_part;
+step s2_auth: SET ROLE regress_cluster_part; SET client_min_messages = ERROR;
 step s2_cluster: CLUSTER cluster_part_tab USING cluster_part_ind; <waiting ...>
 step s1_commit: COMMIT;
 step s2_cluster: <... completed>
@@ -11,7 +11,7 @@ step s2_reset: RESET ROLE;
 
 starting permutation: s1_begin s2_auth s1_lock_parent s2_cluster s1_commit s2_reset
 step s1_begin: BEGIN;
-step s2_auth: SET ROLE regress_cluster_part;
+step s2_auth: SET ROLE regress_cluster_part; SET client_min_messages = ERROR;
 step s1_lock_parent: LOCK cluster_part_tab IN SHARE UPDATE EXCLUSIVE MODE;
 step s2_cluster: CLUSTER cluster_part_tab USING cluster_part_ind; <waiting ...>
 step s1_commit: COMMIT;
@@ -21,17 +21,15 @@ step s2_reset: RESET ROLE;
 starting permutation: s1_begin s1_lock_child s2_auth s2_cluster s1_commit s2_reset
 step s1_begin: BEGIN;
 step s1_lock_child: LOCK cluster_part_tab1 IN SHARE UPDATE EXCLUSIVE MODE;
-step s2_auth: SET ROLE regress_cluster_part;
-step s2_cluster: CLUSTER cluster_part_tab USING cluster_part_ind; <waiting ...>
+step s2_auth: SET ROLE regress_cluster_part; SET client_min_messages = ERROR;
+step s2_cluster: CLUSTER cluster_part_tab USING cluster_part_ind;
 step s1_commit: COMMIT;
-step s2_cluster: <... completed>
 step s2_reset: RESET ROLE;
 
 starting permutation: s1_begin s2_auth s1_lock_child s2_cluster s1_commit s2_reset
 step s1_begin: BEGIN;
-step s2_auth: SET ROLE regress_cluster_part;
+step s2_auth: SET ROLE regress_cluster_part; SET client_min_messages = ERROR;
 step s1_lock_child: LOCK cluster_part_tab1 IN SHARE UPDATE EXCLUSIVE MODE;
-step s2_cluster: CLUSTER cluster_part_tab USING cluster_part_ind; <waiting ...>
+step s2_cluster: CLUSTER cluster_part_tab USING cluster_part_ind;
 step s1_commit: COMMIT;
-step s2_cluster: <... completed>
 step s2_reset: RESET ROLE;
diff --git a/src/test/isolation/specs/cluster-conflict-partition.spec b/src/test/isolation/specs/cluster-conflict-partition.spec
index ae38cb4ee3..4d38a7f49a 100644
--- a/src/test/isolation/specs/cluster-conflict-partition.spec
+++ b/src/test/isolation/specs/cluster-conflict-partition.spec
@@ -23,12 +23,15 @@ step s1_lock_child     { LOCK cluster_part_tab1 IN SHARE UPDATE EXCLUSIVE MODE;
 step s1_commit         { COMMIT; }
 
 session s2
-step s2_auth           { SET ROLE regress_cluster_part; }
+step s2_auth           { SET ROLE regress_cluster_part; SET client_min_messages = ERROR; }
 step s2_cluster        { CLUSTER cluster_part_tab USING cluster_part_ind; }
 step s2_reset          { RESET ROLE; }
 
-# CLUSTER waits if locked, passes for all cases.
+# CLUSTER on the parent waits if locked, passes for all cases.
 permutation s1_begin s1_lock_parent s2_auth s2_cluster s1_commit s2_reset
 permutation s1_begin s2_auth s1_lock_parent s2_cluster s1_commit s2_reset
+
+# When taking a lock on a partition leaf, CLUSTER on the parent skips
+# the leaf, passes for all cases.
 permutation s1_begin s1_lock_child s2_auth s2_cluster s1_commit s2_reset
 permutation s1_begin s2_auth s1_lock_child s2_cluster s1_commit s2_reset
diff --git a/src/test/regress/expected/cluster.out b/src/test/regress/expected/cluster.out
index 2eec483eaa..27a5dff5d4 100644
--- a/src/test/regress/expected/cluster.out
+++ b/src/test/regress/expected/cluster.out
@@ -508,6 +508,7 @@ CREATE TEMP TABLE ptnowner_oldnodes AS
   JOIN pg_class AS c ON c.oid=tree.relid;
 SET SESSION AUTHORIZATION regress_ptnowner;
 CLUSTER ptnowner USING ptnowner_i_idx;
+WARNING:  permission denied to cluster "ptnowner2", skipping it
 RESET SESSION AUTHORIZATION;
 SELECT a.relname, a.relfilenode=b.relfilenode FROM pg_class a
   JOIN ptnowner_oldnodes b USING (oid) ORDER BY a.relname COLLATE "C";
@@ -515,7 +516,7 @@ SELECT a.relname, a.relfilenode=b.relfilenode FROM pg_class a
 -----------+----------
  ptnowner  | t
  ptnowner1 | f
- ptnowner2 | f
+ ptnowner2 | t
 (3 rows)
 
 DROP TABLE ptnowner;
diff --git a/src/test/regress/expected/vacuum.out b/src/test/regress/expected/vacuum.out
index 41e020cf20..4def90b805 100644
--- a/src/test/regress/expected/vacuum.out
+++ b/src/test/regress/expected/vacuum.out
@@ -442,14 +442,20 @@ ALTER TABLE vacowned_parted OWNER TO regress_vacuum;
 ALTER TABLE vacowned_part1 OWNER TO regress_vacuum;
 SET ROLE regress_vacuum;
 VACUUM vacowned_parted;
+WARNING:  permission denied to vacuum "vacowned_part2", skipping it
 VACUUM vacowned_part1;
 VACUUM vacowned_part2;
+WARNING:  permission denied to vacuum "vacowned_part2", skipping it
 ANALYZE vacowned_parted;
+WARNING:  permission denied to analyze "vacowned_part2", skipping it
 ANALYZE vacowned_part1;
 ANALYZE vacowned_part2;
+WARNING:  permission denied to analyze "vacowned_part2", skipping it
 VACUUM (ANALYZE) vacowned_parted;
+WARNING:  permission denied to vacuum "vacowned_part2", skipping it
 VACUUM (ANALYZE) vacowned_part1;
 VACUUM (ANALYZE) vacowned_part2;
+WARNING:  permission denied to vacuum "vacowned_part2", skipping it
 RESET ROLE;
 -- Only one partition owned by other user.
 ALTER TABLE vacowned_parted OWNER TO CURRENT_USER;
@@ -478,14 +484,26 @@ ALTER TABLE vacowned_parted OWNER TO regress_vacuum;
 ALTER TABLE vacowned_part1 OWNER TO CURRENT_USER;
 SET ROLE regress_vacuum;
 VACUUM vacowned_parted;
+WARNING:  permission denied to vacuum "vacowned_part1", skipping it
+WARNING:  permission denied to vacuum "vacowned_part2", skipping it
 VACUUM vacowned_part1;
+WARNING:  permission denied to vacuum "vacowned_part1", skipping it
 VACUUM vacowned_part2;
+WARNING:  permission denied to vacuum "vacowned_part2", skipping it
 ANALYZE vacowned_parted;
+WARNING:  permission denied to analyze "vacowned_part1", skipping it
+WARNING:  permission denied to analyze "vacowned_part2", skipping it
 ANALYZE vacowned_part1;
+WARNING:  permission denied to analyze "vacowned_part1", skipping it
 ANALYZE vacowned_part2;
+WARNING:  permission denied to analyze "vacowned_part2", skipping it
 VACUUM (ANALYZE) vacowned_parted;
+WARNING:  permission denied to vacuum "vacowned_part1", skipping it
+WARNING:  permission denied to vacuum "vacowned_part2", skipping it
 VACUUM (ANALYZE) vacowned_part1;
+WARNING:  permission denied to vacuum "vacowned_part1", skipping it
 VACUUM (ANALYZE) vacowned_part2;
+WARNING:  permission denied to vacuum "vacowned_part2", skipping it
 RESET ROLE;
 DROP TABLE vacowned;
 DROP TABLE vacowned_parted;
-- 
2.25.1


--a8Wt8u1KmwUX3Y2C
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v1-0002-convert-boolean-parameter-into-a-VACOPT_-option.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v48 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a514053b777..781a471fc1d 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us to
+	 * error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3288,9 +3304,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..c20ac682b0d 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some
+					 * process running REPACK (CONCURRENTLY), just fail.  That
+					 * process is going to upgrade its lock at some point, and
+					 * it would be inappropriate for any other process to
+					 * cause that to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+										  MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--qfkt2ktdpcfeypib
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v48-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v46 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a58277bb6c1..21ab81c6f7f 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3123,10 +3130,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3346,9 +3361,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--mdsv5bdgfjjj6d77
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v46-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v47 8/9] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index 9c807f75d71..3acb2e1d2ba 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3287,9 +3303,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--2fxmamo6mu2qbxgv
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v47-0009-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v45 7/8] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index d25ab43da7b..7291b4180ab 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3119,10 +3126,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3342,9 +3357,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--cldir6umqisr673d
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v45-0008-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v46 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a58277bb6c1..21ab81c6f7f 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3123,10 +3130,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3346,9 +3361,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--mdsv5bdgfjjj6d77
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v46-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v47 8/9] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index 9c807f75d71..3acb2e1d2ba 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3287,9 +3303,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--2fxmamo6mu2qbxgv
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v47-0009-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v45 7/8] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index d25ab43da7b..7291b4180ab 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3119,10 +3126,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3342,9 +3357,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--cldir6umqisr673d
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v45-0008-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v46 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a58277bb6c1..21ab81c6f7f 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3123,10 +3130,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3346,9 +3361,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--mdsv5bdgfjjj6d77
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v46-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v47 8/9] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index 9c807f75d71..3acb2e1d2ba 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3287,9 +3303,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--2fxmamo6mu2qbxgv
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v47-0009-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v48 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a514053b777..781a471fc1d 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us to
+	 * error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3288,9 +3304,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..c20ac682b0d 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some
+					 * process running REPACK (CONCURRENTLY), just fail.  That
+					 * process is going to upgrade its lock at some point, and
+					 * it would be inappropriate for any other process to
+					 * cause that to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+										  MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--qfkt2ktdpcfeypib
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v48-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v45 7/8] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index d25ab43da7b..7291b4180ab 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3119,10 +3126,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3342,9 +3357,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--cldir6umqisr673d
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v45-0008-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v46 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a58277bb6c1..21ab81c6f7f 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3123,10 +3130,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3346,9 +3361,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--mdsv5bdgfjjj6d77
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v46-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v47 8/9] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index 9c807f75d71..3acb2e1d2ba 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3287,9 +3303,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--2fxmamo6mu2qbxgv
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v47-0009-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v48 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a514053b777..781a471fc1d 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us to
+	 * error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3288,9 +3304,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..c20ac682b0d 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some
+					 * process running REPACK (CONCURRENTLY), just fail.  That
+					 * process is going to upgrade its lock at some point, and
+					 * it would be inappropriate for any other process to
+					 * cause that to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+										  MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--qfkt2ktdpcfeypib
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v48-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v45 7/8] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index d25ab43da7b..7291b4180ab 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3119,10 +3126,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3342,9 +3357,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--cldir6umqisr673d
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v45-0008-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v46 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a58277bb6c1..21ab81c6f7f 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3123,10 +3130,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3346,9 +3361,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--mdsv5bdgfjjj6d77
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v46-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v47 8/9] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index 9c807f75d71..3acb2e1d2ba 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3287,9 +3303,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--2fxmamo6mu2qbxgv
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v47-0009-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v48 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a514053b777..781a471fc1d 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us to
+	 * error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3288,9 +3304,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..c20ac682b0d 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some
+					 * process running REPACK (CONCURRENTLY), just fail.  That
+					 * process is going to upgrade its lock at some point, and
+					 * it would be inappropriate for any other process to
+					 * cause that to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+										  MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--qfkt2ktdpcfeypib
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v48-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v45 7/8] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index d25ab43da7b..7291b4180ab 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3119,10 +3126,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3342,9 +3357,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--cldir6umqisr673d
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v45-0008-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v46 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a58277bb6c1..21ab81c6f7f 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3123,10 +3130,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3346,9 +3361,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--mdsv5bdgfjjj6d77
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v46-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v47 8/9] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index 9c807f75d71..3acb2e1d2ba 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3287,9 +3303,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--2fxmamo6mu2qbxgv
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v47-0009-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v48 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a514053b777..781a471fc1d 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us to
+	 * error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3288,9 +3304,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..c20ac682b0d 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some
+					 * process running REPACK (CONCURRENTLY), just fail.  That
+					 * process is going to upgrade its lock at some point, and
+					 * it would be inappropriate for any other process to
+					 * cause that to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+										  MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--qfkt2ktdpcfeypib
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v48-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v45 7/8] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index d25ab43da7b..7291b4180ab 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3119,10 +3126,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3342,9 +3357,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--cldir6umqisr673d
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v45-0008-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v46 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a58277bb6c1..21ab81c6f7f 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3123,10 +3130,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3346,9 +3361,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--mdsv5bdgfjjj6d77
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v46-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v47 8/9] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index 9c807f75d71..3acb2e1d2ba 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3287,9 +3303,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--2fxmamo6mu2qbxgv
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v47-0009-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v48 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 52 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 19 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a514053b777..781a471fc1d 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us to
+	 * error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3090,7 +3097,16 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
 	foreach_ptr(RelationData, index, indexrels)
@@ -3288,9 +3304,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..c20ac682b0d 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some
+					 * process running REPACK (CONCURRENTLY), just fail.  That
+					 * process is going to upgrade its lock at some point, and
+					 * it would be inappropriate for any other process to
+					 * cause that to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+										  MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--qfkt2ktdpcfeypib
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v48-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v45 7/8] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index d25ab43da7b..7291b4180ab 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3119,10 +3126,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3342,9 +3357,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--cldir6umqisr673d
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v45-0008-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread

* [PATCH v46 6/7] Error out any process that would block at REPACK
@ 2026-03-25 19:35  Álvaro Herrera <[email protected]>
  0 siblings, 0 replies; 1055+ messages in thread

From: Álvaro Herrera @ 2026-03-25 19:35 UTC (permalink / raw)

Any process waiting on REPACK to release its lock would actually cause
it to deadlock when it tries to upgrade its lock to AEL, losing all work
done to that point.  We avoid this by teaching the deadlock detector to
raise an error when this condition is detected.
---
 src/backend/commands/repack.c                 | 53 ++++++++----
 src/backend/storage/lmgr/deadlock.c           | 15 ++++
 src/include/storage/proc.h                    |  6 +-
 src/test/modules/injection_points/Makefile    |  1 +
 .../expected/repack_deadlock.out              | 63 ++++++++++++++
 src/test/modules/injection_points/meson.build |  1 +
 .../specs/repack_deadlock.spec                | 83 +++++++++++++++++++
 7 files changed, 202 insertions(+), 20 deletions(-)
 create mode 100644 src/test/modules/injection_points/expected/repack_deadlock.out
 create mode 100644 src/test/modules/injection_points/specs/repack_deadlock.spec

diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index a58277bb6c1..21ab81c6f7f 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -276,6 +276,16 @@ ExecRepack(ParseState *pstate, RepackStmt *stmt, bool isTopLevel)
 	/* Determine the lock mode to use. */
 	lockmode = RepackLockLevel((params.options & CLUOPT_CONCURRENT) != 0);
 
+	/*
+	 * If in concurrent mode, set the PROC_IN_CONCURRENT_REPACK flag.  This
+	 * makes the deadlock checker cause anyone that would conflict with us
+	 * to error out.  It's important to set this flag ahead of actually locking
+	 * the relation; it won't of course affect anyone until we do have a lock
+	 * that others can conflict with.
+	 */
+	if ((params.options & CLUOPT_CONCURRENT) != 0)
+		MyProc->statusFlags |= PROC_IN_CONCURRENT_REPACK;
+
 	/*
 	 * If a single relation is specified, process it and we're done ... unless
 	 * the relation is a partitioned table, in which case we fall through.
@@ -476,11 +486,8 @@ RepackLockLevel(bool concurrent)
  * If indexOid is InvalidOid, the table will be rewritten in physical order
  * instead of index order.
  *
- * Note that, in the concurrent case, the function releases the lock at some
- * point, in order to get AccessExclusiveLock for the final steps (i.e. to
- * swap the relation files). To make things simpler, the caller should expect
- * OldHeap to be closed on return, regardless CLUOPT_CONCURRENT. (The
- * AccessExclusiveLock is kept till the end of the transaction.)
+ * On return, OldHeap is closed but locked with AccessExclusiveLock - the lock
+ * will be released at end of the transaction.
  *
  * 'cmd' indicates which command is being executed, to be used for error
  * messages.
@@ -512,10 +519,12 @@ cluster_rel(RepackCommand cmd, Relation OldHeap, Oid indexOid,
 		/*
 		 * Make sure we're not in a transaction block.
 		 *
-		 * The reason is that repack_setup_logical_decoding() could deadlock
-		 * if there's an XID already assigned.  It would be possible to run in
-		 * a transaction block if we had no XID, but this restriction is
-		 * simpler for users to understand and we don't lose anything.
+		 * The reason is that repack_setup_logical_decoding() could wait
+		 * indefinitely for our XID to complete. (The deadlock detector would
+		 * not recognize it because we'd be waiting for ourselves, i.e. no
+		 * real lock conflict.) It would be possible to run in a transaction
+		 * block if we had no XID, but this restriction is simpler for users
+		 * to understand and we don't lose anything.
 		 */
 		PreventInTransactionBlock(isTopLevel, "REPACK (CONCURRENTLY)");
 
@@ -998,10 +1007,8 @@ rebuild_relation(Relation OldHeap, Relation index, bool verbose,
 		 * Note that the worker has to wait for all transactions with XID
 		 * already assigned to finish. If some of those transactions is
 		 * waiting for a lock conflicting with ShareUpdateExclusiveLock on our
-		 * table (e.g.  it runs CREATE INDEX), we can end up in a deadlock.
-		 * Not sure this risk is worth unlocking/locking the table (and its
-		 * clustering index) and checking again if it's still eligible for
-		 * REPACK CONCURRENTLY.
+		 * table (e.g. it runs CREATE INDEX), it should encounter ERROR in the
+		 * deadlock checking code.
 		 */
 		start_repack_decoding_worker(tableOid);
 
@@ -3123,10 +3130,18 @@ rebuild_relation_finish_concurrent(Relation NewHeap, Relation OldHeap,
 		LockRelationOid(OldHeap->rd_rel->reltoastrelid, AccessExclusiveLock);
 
 	/*
-	 * Tuples and pages of the old heap will be gone, but the heap will stay.
+	 * Now that we have all access-exclusive locks on all relations, we no
+	 * longer want other processes to error out when trying to acquire a
+	 * conflicting lock.  Therefore, unset our flag.
+	 */
+	MyProc->statusFlags &= ~PROC_IN_CONCURRENT_REPACK;
+
+	/*
+	 * Tuples and pages of the old heap will be gone, but the heap itself will
+	 * stay.  In order for predicate locks to continue to work, convert them
+	 * to relation-level locks.  We do this both for table and indexes.
 	 */
 	TransferPredicateLocksToHeapRelation(OldHeap);
-	/* The same for indexes. */
 	for (int i = 0; i < nind; i++)
 	{
 		Relation	index = ind_refs[i];
@@ -3346,9 +3361,11 @@ start_repack_decoding_worker(Oid relid)
 
 	/*
 	 * The decoding setup must be done before the caller can have XID assigned
-	 * for any reason, otherwise the worker might end up in a deadlock,
-	 * waiting for the caller's transaction to end. Therefore wait here until
-	 * the worker indicates that it has the logical decoding initialized.
+	 * for any reason, otherwise the worker might end up waiting for the
+	 * caller's transaction to end. (Deadlock detector does not consider this
+	 * a conflict because the worker is in the same locking group as the
+	 * backend that launched it.) Therefore wait here until the worker
+	 * indicates that it has the logical decoding initialized.
 	 */
 	ConditionVariablePrepareToSleep(&shared->cv);
 	for (;;)
diff --git a/src/backend/storage/lmgr/deadlock.c b/src/backend/storage/lmgr/deadlock.c
index b8962d875b6..3d3ec0da111 100644
--- a/src/backend/storage/lmgr/deadlock.c
+++ b/src/backend/storage/lmgr/deadlock.c
@@ -620,6 +620,21 @@ FindLockCycleRecurseMember(PGPROC *checkProc,
 						proc->statusFlags & PROC_IS_AUTOVACUUM)
 						blocking_autovacuum_proc = proc;
 
+					/*
+					 * Similarly, if we note that we're blocked by some process
+					 * running REPACK (CONCURRENTLY), just fail.  That process
+					 * is going to upgrade its lock at some point, and it would
+					 * be inappropriate for any other process to cause that
+					 * to fail.
+					 */
+					if (checkProc == MyProc &&
+						proc->statusFlags & PROC_IN_CONCURRENT_REPACK)
+						ereport(ERROR,
+								errcode(ERRCODE_OBJECT_IN_USE),
+								errmsg("could not wait for concurrent REPACK"),
+								errdetail("Process %d waits for REPACK running on process %d",
+									   MyProc->pid, proc->pid));
+
 					/* We're done looking at this proclock */
 					break;
 				}
diff --git a/src/include/storage/proc.h b/src/include/storage/proc.h
index 1dad125706e..8ad9718f3d6 100644
--- a/src/include/storage/proc.h
+++ b/src/include/storage/proc.h
@@ -69,10 +69,12 @@ struct XidCache
 #define		PROC_AFFECTS_ALL_HORIZONS	0x20	/* this proc's xmin must be
 												 * included in vacuum horizons
 												 * in all databases */
+#define		PROC_IN_CONCURRENT_REPACK	0x40	/* REPACK (CONCURRENTLY) */
 
-/* flags reset at EOXact */
+/* flags reset at EOXact.  A bit of a misnomer ... */
 #define		PROC_VACUUM_STATE_MASK \
-	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND)
+	(PROC_IN_VACUUM | PROC_IN_SAFE_IC | PROC_VACUUM_FOR_WRAPAROUND | \
+	 PROC_IN_CONCURRENT_REPACK)
 
 /*
  * Xmin-related flags. Make sure any flags that affect how the process' Xmin
diff --git a/src/test/modules/injection_points/Makefile b/src/test/modules/injection_points/Makefile
index 2cd7d87c533..f7663859fe2 100644
--- a/src/test/modules/injection_points/Makefile
+++ b/src/test/modules/injection_points/Makefile
@@ -15,6 +15,7 @@ REGRESS_OPTS = --dlpath=$(top_builddir)/src/test/regress
 ISOLATION = basic \
 	    inplace \
 	    repack \
+	    repack_deadlock \
 	    repack_toast \
 	    syscache-update-pruned \
 	    heap_lock_update
diff --git a/src/test/modules/injection_points/expected/repack_deadlock.out b/src/test/modules/injection_points/expected/repack_deadlock.out
new file mode 100644
index 00000000000..a86e4767536
--- /dev/null
+++ b/src/test/modules/injection_points/expected/repack_deadlock.out
@@ -0,0 +1,63 @@
+Parsed test spec with 2 sessions
+
+starting permutation: wait_before_lock add_column wakeup_before_lock check1
+injection_points_attach
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: 
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+ <waiting ...>
+step add_column: 
+	alter table repack_deadlock add column noise text;
+ <waiting ...>
+step add_column: <... completed>
+ERROR:  could not wait for concurrent REPACK
+step wakeup_before_lock: 
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+
+injection_points_wakeup
+-----------------------
+                       
+(1 row)
+
+step wait_before_lock: <... completed>
+step check1: 
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+
+count
+-----
+    1
+(1 row)
+
+i|j
+-+-
+1|1
+2|2
+3|3
+4|4
+(4 rows)
+
+count
+-----
+    4
+(1 row)
+
+injection_points_detach
+-----------------------
+                       
+(1 row)
+
diff --git a/src/test/modules/injection_points/meson.build b/src/test/modules/injection_points/meson.build
index a414abb924b..1cd88d6db65 100644
--- a/src/test/modules/injection_points/meson.build
+++ b/src/test/modules/injection_points/meson.build
@@ -46,6 +46,7 @@ tests += {
       'basic',
       'inplace',
       'repack',
+      'repack_deadlock',
       'repack_toast',
       'syscache-update-pruned',
       'heap_lock_update',
diff --git a/src/test/modules/injection_points/specs/repack_deadlock.spec b/src/test/modules/injection_points/specs/repack_deadlock.spec
new file mode 100644
index 00000000000..9d23a6588c2
--- /dev/null
+++ b/src/test/modules/injection_points/specs/repack_deadlock.spec
@@ -0,0 +1,83 @@
+# Test REPACK with a concurrent transaction that would cause a deadlock
+setup
+{
+	CREATE EXTENSION injection_points;
+
+	CREATE TABLE repack_deadlock(i int PRIMARY KEY, j int);
+	INSERT INTO repack_deadlock(i, j) VALUES (1, 1), (2, 2), (3, 3), (4, 4);
+
+	CREATE TABLE relfilenodes(node oid);
+
+	CREATE TABLE data_s1(i int, j int);
+	CREATE TABLE data_s2(i int, j int);
+}
+
+teardown
+{
+	DROP TABLE repack_deadlock;
+	DROP EXTENSION injection_points;
+
+	DROP TABLE relfilenodes;
+	DROP TABLE data_s1;
+	DROP TABLE data_s2;
+}
+
+session s1
+setup
+{
+	SELECT injection_points_set_local();
+	SELECT injection_points_attach('repack-concurrently-before-lock', 'wait');
+}
+# Perform the initial load and wait for s2 to do some data changes.
+step wait_before_lock
+{
+	REPACK (CONCURRENTLY) repack_deadlock USING INDEX repack_deadlock_pkey;
+}
+# Check the table from the perspective of s1.
+#
+# Besides the contents, we also check that relfilenode has changed.
+
+# Have each session write the contents into a table and use FULL JOIN to check
+# if the outputs are identical.
+step check1
+{
+	INSERT INTO relfilenodes(node)
+	SELECT relfilenode FROM pg_class WHERE relname='repack_deadlock';
+
+	SELECT count(DISTINCT node) FROM relfilenodes;
+
+	SELECT i, j FROM repack_deadlock ORDER BY i, j;
+
+	INSERT INTO data_s1(i, j)
+	SELECT i, j FROM repack_deadlock;
+
+	SELECT count(*)
+	FROM data_s1 d1 FULL JOIN data_s2 d2 USING (i, j)
+	WHERE d1.i ISNULL OR d2.i ISNULL;
+}
+teardown
+{
+	SELECT injection_points_detach('repack-concurrently-before-lock');
+}
+
+session s2
+# Change the existing data. UPDATE changes both key and non-key columns. Also
+# update one row twice to test whether tuple version generated by this session
+# can be found.
+step add_column
+{
+	alter table repack_deadlock add column noise text;
+}
+
+step wakeup_before_lock
+{
+	SELECT injection_points_wakeup('repack-concurrently-before-lock');
+}
+
+# Test if data changes introduced while one session is performing REPACK
+# CONCURRENTLY find their way into the table.
+permutation
+	wait_before_lock
+	add_column
+	wakeup_before_lock
+	check1
-- 
2.47.3


--mdsv5bdgfjjj6d77
Content-Type: text/x-diff; charset=utf-8
Content-Disposition: attachment;
	filename="v46-0007-Teach-snapshot-builder-to-skip-transactions-runn.patch"



^ permalink  raw  reply  [nested|flat] 1055+ messages in thread