RE: BUG #16754: When using LLVM and parallel queries aborted all session by pg_cancel_backend.

Hi,

Thanks for the comment.
I newly cloned the latest version 13.1 source and then rebuilt it. The segmentation fault has recurred.I tried the backtrace feature, but it didn't work. The function specification may not have been good.
Segmentation faults do not always occur, but occasionally after running pg_cancel_backend several times.
Also, it only seems to happen if I canceled the parallel worker process.

Regards,
Noriyoshi Shinoda
-----Original Message-----
From: Andres Freund [mailto:andres(at)anarazel(dot)de]
Sent: Friday, December 4, 2020 5:51 AM
To: Shinoda, Noriyoshi (PN Japan FSIP) <noriyoshi(dot)shinoda(at)hpe(dot)com>
Cc: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>; pgsql-bugs(at)lists(dot)postgresql(dot)org
Subject: Re: BUG #16754: When using LLVM and parallel queries aborted all session by pg_cancel_backend.

Hi,

On 2020-12-02 01:11:07 +0000, Shinoda, Noriyoshi (PN Japan FSIP) wrote:
> > I don't see any plausibly-related commits since 13.1, so it's not something we recently fixed. I wonder if this is an LLVM bug fixed since the 7.1 release you have.
>
> Thank you for your comment and test.
> I tried again with an older version of LLVM below.
>
> - PostgreSQL 13.1
> - Red Hat Enterprise Linux 7.5
> - LLVM 6.0
> - cfe 6.0

Could you try to reproduce the issue with the version you hit it earlier? It'd be helpful to get a backtrace.

> I couldn't reproduce the problem int the above environment, so maybe it's a problem in my development environment.

Did you have any modifications in the tree?

Did you do a build, pulled some updates, and built again? I noticed that you didn't pass --enable-depend to ./configure, so it's possible that not all files got rebuilt that needed to.

Regards,

Andres