Message-ID: <f0f5ab8f-5e3c-edd1-d6ab-d23263759f53@gmail.com>
Date: Sat, 22 Jul 2023 13:00:00 +0300
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.4.2
Subject: Re: BUG #18031: Segmentation fault after deadlock within VACUUM's
 parallel worker
Content-Language: en-US
To: Masahiko Sawada <sawada.mshk@gmail.com>, pgsql-bugs@lists.postgresql.org
Cc: Andres Freund <andres@anarazel.de>
References: <18031-ebe2d08cb405f6cc@postgresql.org>
 <CAD21AoDs7vzK7NErse7xTruqY-FXmM+3K00SdBtMcQhiRNkoeQ@mail.gmail.com>
From: Alexander Lakhin <exclusion@gmail.com>
In-Reply-To: 
 <CAD21AoDs7vzK7NErse7xTruqY-FXmM+3K00SdBtMcQhiRNkoeQ@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Archived-At: 
 <https://www.postgresql.org/message-id/f0f5ab8f-5e3c-edd1-d6ab-d23263759f53%40gmail.com>
Precedence: bulk

Hello Sawada-san,

21.07.2023 11:01, Masahiko Sawada wrote:
> ...
> leading to calling RemoveFromWaitQueue() again. I think we should use
> dclist_delete_from_thoroughly() instead. With the attached patch, the
> issue doesn't happen in my environment.

Thank you for the analysis and the patch! It fixes the issue for me too.

> Another thing I noticed is that the Assert(waitLock) in
> RemoveFromWaitQueue() is useless actually, since we access *waitLock
> before that:

Yeah, I also noticed that, and a simple grep-based script with a little
eyeballing discovered one similar Assert usage:

void
llvm_split_symbol_name(const char *name, char **modname, char **funcname)
{
...
         *funcname = rindex(name, '.');
         (*funcname)++;            /* jump over . */

         *modname = pnstrdup(name + strlen("pgextern."),
                             *funcname - name - strlen("pgextern.") - 1);
         Assert(funcname);

May be it's worth to fix it in passing.

Best regards,
Alexander