From c359178350ac7790437feb817c1fb523838abbd9 Mon Sep 17 00:00:00 2001 From: Tom Lane Date: Thu, 22 Sep 2016 11:34:44 -0400 Subject: Be sure to rewind the tuplestore read pointer in non-leader CTEScan nodes. ExecInitCteScan supposed that it didn't have to do anything to the extra tuplestore read pointer it gets from tuplestore_alloc_read_pointer. However, it needs this read pointer to be positioned at the start of the tuplestore, while tuplestore_alloc_read_pointer is actually defined as cloning the current position of read pointer 0. In normal situations that accidentally works because we initialize the whole plan tree at once, before anything gets read. But it fails in an EvalPlanQual recheck, as illustrated in bug #14328 from Dima Pavlov. To fix, just forcibly rewind the pointer after tuplestore_alloc_read_pointer. The cost of doing so is negligible unless the tuplestore is already in TSS_READFILE state, which wouldn't happen in normal cases. We could consider altering tuplestore's API to make that case cheaper, but that would make for a more invasive back-patch and it doesn't seem worth it. This has been broken probably for as long as we've had CTEs, so back-patch to all supported branches. Discussion: <32468.1474548308@sss.pgh.pa.us> --- src/backend/executor/nodeCtescan.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'src/backend/executor') diff --git a/src/backend/executor/nodeCtescan.c b/src/backend/executor/nodeCtescan.c index 75c1ab3e8aa..56f141c5874 100644 --- a/src/backend/executor/nodeCtescan.c +++ b/src/backend/executor/nodeCtescan.c @@ -224,9 +224,13 @@ ExecInitCteScan(CteScan *node, EState *estate, int eflags) { /* Not the leader */ Assert(IsA(scanstate->leader, CteScanState)); + /* Create my own read pointer, and ensure it is at start */ scanstate->readptr = tuplestore_alloc_read_pointer(scanstate->leader->cte_table, scanstate->eflags); + tuplestore_select_read_pointer(scanstate->leader->cte_table, + scanstate->readptr); + tuplestore_rescan(scanstate->leader->cte_table); } /* -- cgit v1.2.3