diff options
| author | Noah Misch <noah@leadboat.com> | 2024-12-28 07:16:22 -0800 |
|---|---|---|
| committer | Noah Misch <noah@leadboat.com> | 2024-12-28 07:16:27 -0800 |
| commit | 536acda0bc7900c407c76c6ca6293e0ac2041f8b (patch) | |
| tree | 667da1a6e9dcac1452e8f83b8d2540e21c3c80f2 /src/backend/commands/alter.c | |
| parent | cfd6cbcf9be078fbdf9587014231a5772039b386 (diff) | |
In REASSIGN OWNED of a database, lock the tuple as mandated.
Commit aac2c9b4fde889d13f859c233c2523345e72d32b mandated such locking
and attempted to fulfill that mandate, but it missed REASSIGN OWNED.
Hence, it remained possible to lose VACUUM's inplace update of
datfrozenxid if a REASSIGN OWNED processed that database at the same
time. This didn't affect the other inplace-updated catalog, pg_class.
For pg_class, REASSIGN OWNED calls ATExecChangeOwner() instead of the
generic AlterObjectOwner_internal(), and ATExecChangeOwner() fulfills
the locking mandate.
Like in GRANT, implement this by following the locking protocol for any
catalog subject to the generic AlterObjectOwner_internal(). It would
suffice to do this for IsInplaceUpdateOid() catalogs only. Back-patch
to v13 (all supported versions).
Kirill Reshke. Reported by Alexander Kukushkin.
Discussion: https://postgr.es/m/CAFh8B=mpKjAy4Cuun-HP-f_vRzh2HSvYFG3rhVfYbfEBUhBAGg@mail.gmail.com
Diffstat (limited to 'src/backend/commands/alter.c')
| -rw-r--r-- | src/backend/commands/alter.c | 9 |
1 files changed, 8 insertions, 1 deletions
diff --git a/src/backend/commands/alter.c b/src/backend/commands/alter.c index aaae9f482fa..8cd1f5b102b 100644 --- a/src/backend/commands/alter.c +++ b/src/backend/commands/alter.c @@ -60,6 +60,7 @@ #include "miscadmin.h" #include "parser/parse_func.h" #include "rewrite/rewriteDefine.h" +#include "storage/lmgr.h" #include "tcop/utility.h" #include "utils/builtins.h" #include "utils/fmgroids.h" @@ -945,7 +946,9 @@ AlterObjectOwner_internal(Relation rel, Oid objectId, Oid new_ownerId) Oid old_ownerId; Oid namespaceId = InvalidOid; - oldtup = get_catalog_object_by_oid(rel, Anum_oid, objectId); + /* Search tuple and lock it. */ + oldtup = + get_catalog_object_by_oid_extended(rel, Anum_oid, objectId, true); if (oldtup == NULL) elog(ERROR, "cache lookup failed for object %u of catalog \"%s\"", objectId, RelationGetRelationName(rel)); @@ -1044,6 +1047,8 @@ AlterObjectOwner_internal(Relation rel, Oid objectId, Oid new_ownerId) /* Perform actual update */ CatalogTupleUpdate(rel, &newtup->t_self, newtup); + UnlockTuple(rel, &oldtup->t_self, InplaceUpdateTupleLock); + /* * Update owner dependency reference. When working on a large object, * we have to translate back to the OID conventionally used for LOs' @@ -1061,6 +1066,8 @@ AlterObjectOwner_internal(Relation rel, Oid objectId, Oid new_ownerId) } else { + UnlockTuple(rel, &oldtup->t_self, InplaceUpdateTupleLock); + /* * No need to change anything. But when working on a large object, we * have to translate back to the OID conventionally used for LOs' |