database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-06-11 20:28:21 +03:00
Code Activity

Fix failure due to accessing an already-freed tuple descriptor in a plan

Browse Source 
involving HashAggregate over SubqueryScan (this is the known case, there
may well be more).  The bug is only latent in releases before 8.2 since they
didn't try to access tupletable slots' descriptors during ExecDropTupleTable.
The least bogus fix seems to be to make subqueries share the parent query's
memory context, so that tupdescs they create will have the same lifespan as
those of the parent query.  There are comments in the code envisioning going
even further by not having a separate child EState at all, but that will
require rethinking executor access to range tables, which I don't want to
tackle right now.  Per bug report from Jean-Pierre Pelletier.
This commit is contained in:
Tom Lane
2006-12-26 21:37:20 +00:00
parent 68996463d4
commit 0cbc5b1ed4
6 changed files with 63 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
src/backend/executor/nodeSubplan.c
View File

@ -7,7 +7,7 @@
* Portions Copyright (c) 1994, Regents of the University of California
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/executor/nodeSubplan.c,v 1.80 2006/10/04 00:29:53 momjian Exp $
* $PostgreSQL: pgsql/src/backend/executor/nodeSubplan.c,v 1.81 2006/12/26 21:37:19 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -635,7 +635,6 @@ ExecInitSubPlan(SubPlanState *node, EState *estate, int eflags)
{
SubPlan *subplan = (SubPlan *) node->xprstate.expr;
EState *sp_estate;
MemoryContext oldcontext;
/*
* Do access checking on the rangetable entries in the subquery.
@ -661,15 +660,13 @@ ExecInitSubPlan(SubPlanState *node, EState *estate, int eflags)
* create an EState for the subplan
*
* The subquery needs its own EState because it has its own rangetable. It
* shares our Param ID space, however. XXX if rangetable access were done
* differently, the subquery could share our EState, which would eliminate
* some thrashing about in this module...
* shares our Param ID space and es_query_cxt, however. XXX if rangetable
* access were done differently, the subquery could share our EState,
* which would eliminate some thrashing about in this module...
*/
sp_estate = CreateExecutorState();
sp_estate = CreateSubExecutorState(estate);
node->sub_estate = sp_estate;
oldcontext = MemoryContextSwitchTo(sp_estate->es_query_cxt);
sp_estate->es_range_table = subplan->rtable;
sp_estate->es_param_list_info = estate->es_param_list_info;
sp_estate->es_param_exec_vals = estate->es_param_exec_vals;
@ -694,8 +691,6 @@ ExecInitSubPlan(SubPlanState *node, EState *estate, int eflags)
node->needShutdown = true; /* now we need to shutdown the subplan */
MemoryContextSwitchTo(oldcontext);
/*
* If this plan is un-correlated or undirect correlated one and want to
* set params for parent plan then mark parameters as needing evaluation.
@ -1036,11 +1031,7 @@ ExecEndSubPlan(SubPlanState *node)
{
if (node->needShutdown)
{
MemoryContext oldcontext;
oldcontext = MemoryContextSwitchTo(node->sub_estate->es_query_cxt);
ExecEndPlan(node->planstate, node->sub_estate);
MemoryContextSwitchTo(oldcontext);
FreeExecutorState(node->sub_estate);
node->sub_estate = NULL;
node->planstate = NULL;