Thanks for your information! I checked it again. Seems this bug can be reproduced when I used "quick-fix-for-bug-17985.patch" for bug 17985, but cannot be reproduced after I used "better-fix-for-bug-17985.patch". So it was fixed by the patch "better-fix-for-bug-17985.patch".
Best wishes,
Zuming
Tomas Vondra <tomas.vondra@enterprisedb.com> writes:
On 6/20/23 22:54, PG Bug reporting form wrote:
My fuzzer finds a correctness bug in Postgres, which makes Postgres return
inconsistent results.
I can't reproduce this with current master - perhaps one of the commits
since efeb12ef0b fixes this too?
It looks like this is another manifestation of the hashjoin bug
fixed in 45392626c.
regards, tom lane