10BC0 Repair failure with SubPlans in multi-row VALUES lists. · koderP/postgres@ae6ed07 · GitHub
[go: up one dir, main page]
Skip to content

Commit ae6ed07

Browse files
tglsfdctglsfdc
committed
Repair failure with SubPlans in multi-row VALUES lists.
When nodeValuesscan.c was written, it was impossible to have a SubPlan in VALUES --- any sub-SELECT there would have to be uncorrelated and thereby would produce an InitPlan instead. We therefore took a shortcut in the logic that throws away a ValuesScan's per-row expression evaluation data structures. This was broken by the introduction of LATERAL however; a sub-SELECT containing a lateral reference produces a correlated SubPlan. The cleanest fix for this would be to give up the optimization of discarding the expression eval state. But that still seems pretty unappetizing for long VALUES lists. It seems to work to just prevent the subexpressions from hooking into the ValuesScan node's subPlan list, so let's do that and see how well it works. (If this breaks, due to additional connections between the subexpressions and the outer query structures, we might consider compromises like throwing away data only for VALUES rows not containing SubPlans.) Per bug #14924 from Christian Duta. Back-patch to 9.3 where LATERAL was introduced. Discussion: https://postgr.es/m/20171124120836.1463.5310@wrigleys.postgresql.org
1 parent 1a164f9 commit ae6ed07

File tree

3 files changed

+56
-5
lines changed

3 files changed

+56
-5
lines changed

‎src/backend/executor/nodeValuesscan.c‎

Lines changed: 16 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -92,6 +92,7 @@ ValuesNext(ValuesScanState *node)
9292
if (exprlist)
9393
{
9494
MemoryContext oldContext;
95+
List *oldsubplans;
9596
List *exprstatelist;
9697
Datum *values;
9798
bool *isnull;
@@ -115,12 +116,22 @@ ValuesNext(ValuesScanState *node)
115116
oldContext = MemoryContextSwitchTo(econtext->ecxt_per_tuple_memory);
116117

117118
/*
118-
* Pass NULL, not my plan node, because we don't want anything in this
119-
* transient state linking into permanent state. The only possibility
120-
* is a SubPlan, and there shouldn't be any (any subselects in the
121-
* VALUES list should be InitPlans).
119+
* The expressions might contain SubPlans (this is currently only
120+
* possible if there's a sub-select containing a LATERAL reference,
121+
* otherwise sub-selects in a VALUES list should be InitPlans). Those
122+
* subplans will want to hook themselves into our subPlan list, which
123+
* would result in a corrupted list after we delete the eval state. We
124+
* can work around this by saving and restoring the subPlan list.
125+
* (There's no need for the functionality that would be enabled by
126+
* having the list entries, since the SubPlans aren't going to be
127+
* re-executed anyway.)
122128
*/
123-
exprstatelist = (List *) ExecInitExpr((Expr *) exprlist, NULL);
129+
oldsubplans = node->ss.ps.subPlan;
130+
node->ss.ps.subPlan = NIL;
131+
132+
exprstatelist = (List *) ExecInitExpr((Expr *) exprlist, &node->ss.ps);
133+
134+
node->ss.ps.subPlan = oldsubplans;
124135

125136
/* parser should have checked all sublists are the same length */
126137
Assert(list_length(exprstatelist) == slot->tts_tupleDescriptor->natts);

‎src/test/regress/expected/subselect.out‎

Lines changed: 30 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -772,6 +772,36 @@ select exists(select * from nocolumns);
772772
f
773773
(1 row)
774774

775+
--
776+
-- Check behavior with a SubPlan in VALUES (bug #14924)
777+
--
778+
select val.x
779+
from generate_series(1,10) as s(i),
780+
lateral (
781+
values ((select s.i + 1)), (s.i + 101)
782+
) as val(x)
783+
where s.i < 10 and (select val.x) < 110;
784+
x
785+
-----
786+
2
787+
102
788+
3
789+
103
790+
4
791+
104
792+
5
793+
105
794+
6
795+
106
796+
7
797+
107
798+
8
799+
108
800+
9
801+
109
802+
10
803+
(17 rows)
804+
775805
--
776806
-- Check sane behavior with nested IN SubLinks
777807
--

‎src/test/regress/sql/subselect.sql‎

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -439,6 +439,16 @@ explain (verbose, costs off)
439439
create temp table nocolumns();
440440
select exists(select * from nocolumns);
441441

442+
--
443+
-- Check behavior with a SubPlan in VALUES (bug #14924)
444+
--
445+
select val.x
446+
from generate_series(1,10) as s(i),
447+
lateral (
448+
values ((select s.i + 1)), (s.i + 101)
449+
) as val(x)
450+
where s.i < 10 and (select val.x) < 110;
451+
442452
--
443453
-- Check sane behavior with nested IN SubLinks
444454
--

0 commit comments

Comments
 (0)
0