Re: Memory consumed by child SpecialJoinInfo in partitionwise join planning

On Fri, Jul 28, 2023 at 7:28 PM Ashutosh Bapat <ashutosh(dot)bapat(dot)oss(at)gmail(dot)com>
wrote:

> >
> > + bms_free(child_sjinfo->commute_above_l);
> > + bms_free(child_sjinfo->commute_above_r);
> > + bms_free(child_sjinfo->commute_below_l);
> > + bms_free(child_sjinfo->commute_below_r);
> >
> > These four members in SpecialJoinInfo only contain outer join relids.
> > They do not need to be translated. So they would reference the same
> > memory areas in child_sjinfo as in parent_sjinfo. We should not free
> > them, otherwise they would become dangling pointers in parent sjinfo.
> >
>
>
Attached patchset fixing those.
0001 - patch to report planning memory, with to explain.out regression
output fix. We may consider committing this as well.
0002 - with your comment addressed above.

>
> I think we should 1. add an assert to make sure that commute_above_*
> do not require any transations i.e. they do not contain any parent
> relids of the child rels.

Looking at the definitions of commute_above and commute_below relids and OJ
relids, I don't think these assertions are necessary. So didn't add those.

> 2. Do not free those.

Done

> 3. Add a comment about
> keeping the build and free functions in sync.
>

Done.

--
Best Wishes,
Ashutosh Bapat