forked frompostgres/postgres
- Notifications
You must be signed in to change notification settings - Fork6
Commite46753e
committed
Fix cache flush hazard in cache_record_field_properties().
We need to increment the refcount on the composite type's cached tupledescriptor while we do lookups of its column types. Otherwise a cacheflush could occur and release the tuple descriptor before we're done withit. This fails reliably with -DCLOBBER_CACHE_ALWAYS, but the odds of afailure in a production build seem rather low (since the pfree'd descriptortypically wouldn't get scribbled on immediately). That may explain thelack of any previous reports. Buildfarm issue noted by Christian Ullrich.Back-patch to 9.1 where the bogus code was added.1 parent219ef8e commite46753e
1 file changed
+5
-0
lines changed| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
578 | 578 | | |
579 | 579 | | |
580 | 580 | | |
| 581 | + | |
| 582 | + | |
| 583 | + | |
581 | 584 | | |
582 | 585 | | |
583 | 586 | | |
| |||
601 | 604 | | |
602 | 605 | | |
603 | 606 | | |
| 607 | + | |
| 608 | + | |
604 | 609 | | |
605 | 610 | | |
606 | 611 | | |
| |||
0 commit comments
Comments
(0)