public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "michalz at nvidia dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c++/99613] New: Static variable destruction order race condition
Date: Tue, 16 Mar 2021 10:58:33 +0000	[thread overview]
Message-ID: <bug-99613-4@http.gcc.gnu.org/bugzilla/> (raw)

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99613

            Bug ID: 99613
           Summary: Static variable destruction order race condition
           Product: gcc
           Version: 7.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: michalz at nvidia dot com
  Target Milestone: ---

Created attachment 50394
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=50394&action=edit
Race condition demo

Function static variables should be destroyed in the reverse order of
initialization. However, the lock is released before calling atexit to register
the destructor, allowing for a race condition to occur if multiple threads are
accessing static variables for the first time.

The bug affects all versions of GCC that I've tried (6.4, 7.3, 8.2, 9.3, 10.2)
on multiple platforms.

The error in output assembly can be seen when compiling the following code:

struct S{
   S();
   ~S();
};

S& foo(){
    static S s;
    return s;
}

The attached file contains a working example that triggers the race condition
(it's a race condition, so sometimes it succeeds).

             reply	other threads:[~2021-03-16 10:58 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-16 10:58 michalz at nvidia dot com [this message]
2021-03-16 11:42 ` [Bug c++/99613] " rguenth at gcc dot gnu.org
2021-03-16 12:23 ` jakub at gcc dot gnu.org
2021-03-16 12:35 ` jakub at gcc dot gnu.org
2021-03-16 12:37 ` michalz at nvidia dot com
2021-03-16 12:42 ` jakub at gcc dot gnu.org
2021-03-16 12:48 ` rguenth at gcc dot gnu.org
2021-03-16 12:56 ` jakub at gcc dot gnu.org
2021-03-16 13:04 ` michalz at nvidia dot com
2021-03-16 13:07 ` michalz at nvidia dot com
2021-03-16 13:15 ` jakub at gcc dot gnu.org
2021-03-16 13:24 ` michalz at nvidia dot com
2021-03-16 13:25 ` michalz at nvidia dot com
2021-03-16 13:35 ` jakub at gcc dot gnu.org
2021-03-16 14:05 ` michalz at nvidia dot com
2021-03-16 14:07 ` jacobhemstad at gmail dot com
2021-03-16 14:23 ` jakub at gcc dot gnu.org
2021-03-16 14:38 ` rguenth at gcc dot gnu.org
2021-03-16 20:18 ` cvs-commit at gcc dot gnu.org
2021-03-19 23:30 ` cvs-commit at gcc dot gnu.org
2021-04-20 23:33 ` cvs-commit at gcc dot gnu.org
2021-04-22 16:51 ` cvs-commit at gcc dot gnu.org
2021-04-22 17:10 ` jakub at gcc dot gnu.org
2021-09-11 14:22 ` pinskia at gcc dot gnu.org

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

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=bug-99613-4@http.gcc.gnu.org/bugzilla/ \
    --to=gcc-bugzilla@gcc.gnu.org \
    --cc=gcc-bugs@gcc.gnu.org \
    /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).