Fix potential SSI hazard in heap_update().
authorThomas Munro <tmunro@postgresql.org>
Tue, 13 Apr 2021 00:34:25 +0000 (12:34 +1200)
committerThomas Munro <tmunro@postgresql.org>
Tue, 13 Apr 2021 01:02:56 +0000 (13:02 +1200)
commitb1df6b696b759f00ebbf02e6de64e259d4be5785
tree1f87911fa68e9749b71cae2524322dbc30ed0c01
parent885a87641930778d9580fdf0656af24e3f52d276
Fix potential SSI hazard in heap_update().

Commit 6f38d4dac38 failed to heed a warning about the stability of the
value pointed to by "otid".  The caller is allowed to pass in a pointer to
newtup->t_self, which will be updated during the execution of the
function.  Instead, the SSI check should use the value we copy into
oldtup.t_self near the top of the function.

Not a live bug, because newtup->t_self doesn't really get updated until
a bit later, but it was confusing and broke the rule established by the
comment.

Back-patch to 13.

Reported-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: http://postgr.es/m/2689164.1618160085%40sss.pgh.pa.us
src/backend/access/heap/heapam.c