Fix brain fade in DefineIndex(): it was continuing to access the table's

relcache entry after having heap_close'd it.  This could lead to misbehavior
if a relcache flush wiped out the cache entry meanwhile.  In 8.2 there is a
very real risk of CREATE INDEX CONCURRENTLY using the wrong relid for locking
and waiting purposes.  I think the bug is only cosmetic in 8.0 and 8.1,
because their transgression is limited to using RelationGetRelationName(rel)
in an ereport message immediately after heap_close, and there's no way (except
with special debugging options) for a cache flush to occur in that interval.
Not quite sure that it's cosmetic in 7.4, but seems best to patch anyway.

Found by trying to run the regression tests with CLOBBER_CACHE_ALWAYS enabled.
Maybe we should try to do that on a regular basis --- it's awfully slow,
but perhaps some fast buildfarm machine could do it once in awhile.

diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index a7471c20dc0a5b5268474284fd3a8b76409cad7e..4d4980748f25977577b9e97647caf75728ffd9f1 100644 (file)
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -112,8 +112,6 @@ DefineIndex(RangeVar *heapRelation,
        relationId = RelationGetRelid(rel);
        namespaceId = RelationGetNamespace(rel);
 
-       heap_close(rel, NoLock);
-
        /*
         * Verify we (still) have CREATE rights in the rel's namespace.
         * (Presumably we did when the rel was created, but maybe not
@@ -251,6 +249,8 @@ DefineIndex(RangeVar *heapRelation,
        ComputeIndexAttrs(indexInfo, classObjectId, attributeList,
                                          relationId, accessMethodName, accessMethodId);
 
+       heap_close(rel, NoLock);
+
        index_create(relationId, indexRelationName,
                                 indexInfo, accessMethodId, classObjectId,
                                 primary, isconstraint, allowSystemTableMods);