Re: BUG #17512: Process running query fails with SIGSEV - nodeMemoize.c:349

Aleš Zelený <zeleny.ales@gmail.com>

From: Aleš Zelený <zeleny.ales@gmail.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: David Rowley <dgrowleyml@gmail.com>, PostgreSQL mailing lists <pgsql-bugs@lists.postgresql.org>
Date: 2022-06-08T20:11:12Z
Lists: pgsql-bugs
Hi all,

is there something, on top of the llvm packages, I shall test or provide to
help the root cause be analyzed?
If I understand the patch properly, I'll avoid SIGSEV, but still the query
with jit=on will fail, except (which is a significant improvement!) the
server will not be restarted.

Thanks Ales

út 7. 6. 2022 v 22:26 odesílatel Aleš Zelený <zeleny.ales@gmail.com> napsal:

> Hello,
>
> here are the installed llvm packages:
>
> llvm-3.4.2-9.el7.x86_64
> llvm5.0-libs-5.0.1-7.el7.x86_64
> llvm5.0-5.0.1-7.el7.x86_64
> llvm-libs-3.4.2-9.el7.x86_64
> postgresql14-llvmjit-14.2-1PGDG.rhel7.x86_64
>
> Kind regards Ales Zeleny
>
> út 7. 6. 2022 v 21:07 odesílatel Tom Lane <tgl@sss.pgh.pa.us> napsal:
>
>> > So far, it looks that the query I've been using to prove the SIGSEV
>> works
>> > fine when I disabled jit by setting
>> > jit=off
>>
>> Oh!  So it's (probably) not a Memoize problem, it's an LLVM problem.
>>
>> > The jit package is postgresql14-llvmjit-14.2-1PGDG.rhel7.x86_64
>>
>> Hm.  What LLVM version have you got on that machine?  (And I wonder
>> whether it's the same as what Devrim built this PGDG package against.)
>>
>>                         regards, tom lane
>>
>

Commits

  1. Harden Memoization code against broken data types