From: ludo@gnu.org (Ludovic Courtès)
To: Andy Wingo <wingo@pobox.com>
Cc: guile-devel@gnu.org
Subject: Re: wip-threads-and-fork
Date: Wed, 21 Mar 2012 22:26:12 +0100 [thread overview]
Message-ID: <8762dx7iiz.fsf@gnu.org> (raw)
In-Reply-To: <87zkbwa95w.fsf@pobox.com> (Andy Wingo's message of "Sun, 04 Mar 2012 12:38:03 +0100")
Hello!
(With delay...)
Andy Wingo <wingo@pobox.com> skribis:
> I would have preferred this, but I came to the conclusion that this
> approach is not sound.
Without exposing ‘pthread_atfork’, how would you suggest making user
code “fork-safe”? A use case would be reviving the futures thread pool
after ‘fork’.
> Did you see that I merged the atfork bits into master?
> (wip-threads-and-fork also had some CLOEXEC bits that needed more
> baking). They worked... sorta. They had a few problems:
>
> 1) It's impossible to work around the lack of atfork() in libraries
> that you depend on.
>
> For example, wip-threads-and-fork called fork() within the GC alloc
> lock, to get around the lack of a pthread_atfork() in libgc. But
> then I submitted a patch to make libgc do this itself:
>
> http://thread.gmane.org/gmane.comp.programming.garbage-collection.boehmgc/4940
>
> It's pretty difficult to tell which version of libgc you would
> have. There is no workaround that is sufficient.
Indeed, good point.
> 2) POSIX explicitly disclaims the result of calling non-signal-safe
> primitives after a fork() of a multithreaded program.
Right, though reality seems to be more pleasant than POSIX. ;-)
> 3) Nobody cares about these bugs. See e.g. the lack of response at
> http://sourceware.org/bugzilla/show_bug.cgi?id=13725. Even Bruno
> didn't reply to the Cc. See point (2).
>
> 4) The atfork mechanism imposes a total ordering on locks. This is
> possible for static locks, but difficult for locks on collectable
> Scheme objects.
>
> 5) Relatedly, just to be able to lock all weak tables at a fork, we
> had to create a new weak table-of-tables and add the tables to it.
> This is needless complication and overhead.
>
> 6) scm_c_atfork() is a broken interface. Because it hangs its hooks
> off of one pthread_atfork() invocation, it can cause newer locks to
> insert themselves in the wrong position relative to
> pthread_atfork() calls made between when Guile installed the
> scm_c_atfork handler, and the call to scm_c_atfork.
>
> There can be only one pthread_atfork() list, in a correct program.
OK, thanks for the nice summary. Indeed, this is a complex story.
> In the end I reverted those patches because they were just complication
> that didn't solve any fundamental problems.
OK.
> I came to the opinion, having run a threaded, forking program, that we
> would be much better off if we provided good abstractions to spawn
> processes, but that expecting fork() to work in a multithreaded program
> is not realistic.
Yes, things like ‘open-process’ make sense.
What about adding a big fat warning in the doc about use of
‘primitive-fork’ in a multi-threaded program?
> Still, there is one other thing that perhaps we could do to shut down
> the signal handling thread around a fork(). Dunno, perhaps it is worth
> looking into.
What would be the expected benefit?
Thanks,
Ludo’.
next prev parent reply other threads:[~2012-03-21 21:26 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-02-08 22:10 wip-threads-and-fork Andy Wingo
2012-02-22 21:40 ` wip-threads-and-fork Andy Wingo
2012-02-23 15:05 ` wip-threads-and-fork Andy Wingo
2012-02-23 15:49 ` wip-threads-and-fork Nala Ginrut
2012-02-23 16:13 ` wip-threads-and-fork Andy Wingo
2012-02-24 3:00 ` wip-threads-and-fork Nala Ginrut
2012-02-24 10:21 ` wip-threads-and-fork Andy Wingo
2012-02-24 14:08 ` wip-threads-and-fork Nala Ginrut
2012-02-24 14:47 ` wip-threads-and-fork Andy Wingo
2012-02-24 15:25 ` wip-threads-and-fork Nala Ginrut
2012-02-26 22:03 ` wip-threads-and-fork Ludovic Courtès
2012-02-27 9:44 ` wip-threads-and-fork Andy Wingo
2012-03-01 19:40 ` wip-threads-and-fork Ludovic Courtès
2012-02-24 18:57 ` wip-threads-and-fork Andy Wingo
2012-02-25 2:21 ` wip-threads-and-fork Nala Ginrut
2012-02-25 2:30 ` wip-threads-and-fork Nala Ginrut
2012-02-25 18:01 ` wip-threads-and-fork Andy Wingo
2012-02-26 2:35 ` wip-threads-and-fork Nala Ginrut
2012-02-26 22:00 ` wip-threads-and-fork Ludovic Courtès
2012-02-27 9:39 ` wip-threads-and-fork Andy Wingo
2012-03-01 19:35 ` wip-threads-and-fork Ludovic Courtès
2012-03-03 16:32 ` wip-threads-and-fork Andy Wingo
2012-03-03 21:20 ` wip-threads-and-fork Ludovic Courtès
2012-03-04 11:38 ` wip-threads-and-fork Andy Wingo
2012-03-21 21:26 ` Ludovic Courtès [this message]
2012-03-22 2:48 ` wip-threads-and-fork Nala Ginrut
2012-03-23 9:40 ` wip-threads-and-fork Andy Wingo
2012-03-27 15:54 ` wip-threads-and-fork Ludovic Courtès
2012-04-06 18:30 ` wip-threads-and-fork Andy Wingo
2012-04-07 22:54 ` wip-threads-and-fork Ludovic Courtès
2013-01-17 11:41 ` wip-threads-and-fork Andy Wingo
2012-03-01 19:32 ` wip-threads-and-fork Ludovic Courtès
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: https://www.gnu.org/software/guile/
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=8762dx7iiz.fsf@gnu.org \
--to=ludo@gnu.org \
--cc=guile-devel@gnu.org \
--cc=wingo@pobox.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).