public inbox for glibc-bugs@sourceware.org help / color / mirror / Atom feed
From: "vitalybuka at google dot com" <sourceware-bugzilla@sourceware.org> To: glibc-bugs@sourceware.org Subject: [Bug libc/27749] New: Data race __run_exit_handlers Date: Sat, 17 Apr 2021 16:15:15 +0000 [thread overview] Message-ID: <bug-27749-131@http.sourceware.org/bugzilla/> (raw) https://sourceware.org/bugzilla/show_bug.cgi?id=27749 Bug ID: 27749 Summary: Data race __run_exit_handlers Product: glibc Version: unspecified Status: UNCONFIRMED Severity: normal Priority: P2 Component: libc Assignee: unassigned at sourceware dot org Reporter: vitalybuka at google dot com CC: drepper.fsp at gmail dot com Target Milestone: --- In our fleet it mostly manifested with tsan builds, which is very sensitive to callback called twice. But I managed to reproduce with code like below of regular builds. With asserts it triggers internal accerts, without them it calls atexit callback twice for the same argument. It very similar to bug 14333, but the fix for that bug just significantly reduced probability of the data race but not eliminated it completely. I reproduced it on 2.27 and head, but I should be reproducible for earlier versions as well. #include <stdatomic.h> #include <stdio.h> #include <stdlib.h> #include <support/xthread.h> #include <sys/wait.h> #include <unistd.h> static atomic_int registered; static atomic_int todo = 100000; static void atexit_cb (void *arg) { --registered; static void *prev; if (arg == prev) { printf ("%p\n", arg); abort (); } prev = arg; while (todo > 0 && registered < 100) ; } int __cxa_atexit (void (*func) (void *), void *arg, void *d); static void *cb_arg = NULL; static void add_handlers (void) { int n = 10; for (int i = 0; i < n; ++i) __cxa_atexit (&atexit_cb, ++cb_arg, 0); registered += n; todo -= n; } static void * thread_func (void *arg) { while (todo > 0) if (registered < 10000) add_handlers (); return 0; } static void test_and_exit (void) { pthread_attr_t attr; xpthread_attr_init (&attr); xpthread_attr_setdetachstate (&attr, 1); xpthread_create (&attr, thread_func, NULL); xpthread_attr_destroy (&attr); while (!registered) ; exit (0); } static int do_test (void) { for (int i = 0; i < 20; ++i) { for (int i = 0; i < 10; ++i) if (fork () == 0) test_and_exit (); int status; while (wait (&status) > 0) { if (!WIFEXITED (status)) { printf ("Failed interation %d\n", i); abort (); } } } exit (0); } #define TEST_FUNCTION do_test #include <support/test-driver.c> -- You are receiving this mail because: You are on the CC list for the bug.
next reply other threads:[~2021-04-17 16:15 UTC|newest] Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top 2021-04-17 16:15 vitalybuka at google dot com [this message] 2021-04-17 17:16 ` [Bug libc/27749] " vitalybuka at google dot com 2021-04-17 17:24 ` vitalybuka at google dot com 2021-05-14 14:38 ` adhemerval.zanella at linaro dot 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-27749-131@http.sourceware.org/bugzilla/ \ --to=sourceware-bugzilla@sourceware.org \ --cc=glibc-bugs@sourceware.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: linkBe 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).