database/postgres
database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-05-08 07:21:33 +03:00
Code

Fix edge case leading to agg transitions skipping ExecAggTransReparent() calls.

Browse Source 
The code checking whether an aggregate transition value needs to be
reparented into the current context has always only compared the
transition return value with the previous transition value by datum,
i.e. without regard for NULLness.  This normally works, because when
the transition function returns NULL (via fcinfo->isnull), it'll
return a value that won't be the same as its input value.

But there's no hard requirement that that's the case. And it turns
out, it's possible to hit this case (see discussion or reproducers),
leading to a non-null transition value not being reparented, followed
by a crash caused by that.

Instead of adding another comparison of NULLness, instead have
ExecAggTransReparent() ensure that pergroup->transValue ends up as 0
when the new transition value is NULL. That avoids having to add an
additional branch to the much more common cases of the transition
function returning the old transition value (which is a pointer in
this case), and when the new value is different, but not NULL.

In branches since 69c3936a149, also deduplicate the reparenting code
between the expression evaluation based transitions, and the path for
ordered aggregates.

Reported-By: Teodor Sigaev, Nikita Glukhov
Author: Andres Freund
Discussion: https://postgr.es/m/bd34e930-cfec-ea9b-3827-a8bc50891393@sigaev.ru
Backpatch: 9.4-, this issue has existed since at least 7.4
This commit is contained in:
Andres Freund 2020-01-20 23:26:51 -08:00
parent 9740cdbe5e
commit d4c339924c
1 changed files with 19 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

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

@ -812,6 +812,15 @@ advance_transition_function(AggState *aggstate,
* first input, we don't need to do anything. Also, if transfn returned a * first input, we don't need to do anything. Also, if transfn returned a
* pointer to a R/W expanded object that is already a child of the * pointer to a R/W expanded object that is already a child of the
* aggcontext, assume we can adopt that value without copying it. * aggcontext, assume we can adopt that value without copying it.
*
* It's safe to compare newVal with pergroupstate->transValue without
* regard for either being NULL, because we below take care to set
* transValue to 0 when NULL. Otherwise we could end up accidentally not
* reparenting, when the transValue has the same numerical value as
* newVal, despite being NULL. This is a somewhat hot path, making it
* undesirable to instead solve this with another branch for the common
* case of the transition function returning its (modified) input
* argument.
*/ */
if (!pertrans->transtypeByVal && if (!pertrans->transtypeByVal &&
DatumGetPointer(newVal) != DatumGetPointer(pergroupstate->transValue)) DatumGetPointer(newVal) != DatumGetPointer(pergroupstate->transValue))
@ -829,6 +838,16 @@ advance_transition_function(AggState *aggstate,
pertrans->transtypeByVal, pertrans->transtypeByVal,
pertrans->transtypeLen); pertrans->transtypeLen);
} }
else
{
/*
* Ensure that pergroupstate->transValue ends up being 0, so we
* can safely compare newVal/transValue without having to check
* their respective nullness.
*/
newVal = (Datum) 0;
}
if (!pergroupstate->transValueIsNull) if (!pergroupstate->transValueIsNull)
{ {
if (DatumIsReadWriteExpandedObject(pergroupstate->transValue, if (DatumIsReadWriteExpandedObject(pergroupstate->transValue,