On Tue, 7 May 2024 at 17:28, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> David Rowley <dgrowleyml@gmail.com> writes:
> > Yeah, before the revert, that did:
> > - sjinf->syn_lefthand = replace_relid(sjinf->syn_lefthand, relid, subst);
> > That replace code seems to have always done a bms_copy()
>
> Hmm, not always; see e0477837c.
It was the discussion on that thread that led to the invention of
REALLOCATE_BITMAPSETS
> What I'm trying to figure out here is whether we have a live bug
> in this area in released branches; and if so, why we've not seen
> reports of that.
We could check what portions of REALLOCATE_BITMAPSETS are
backpatchable. It may not be applicable very far back because of v16's
00b41463c. The bms_del_member() would have left a zero set rather than
doing bms_free() prior to that commit. There could be a bug in v16.
David