public inbox for fortran@gcc.gnu.org
 help / color / mirror / Atom feed
From: Bernd Edlinger <bernd.edlinger@hotmail.de>
To: Janne Blomqvist <blomqvist.janne@gmail.com>
Cc: Thomas Koenig <tkoenig@netcologne.de>,
	"fortran@gcc.gnu.org"	<fortran@gcc.gnu.org>,
	"gcc-patches@gcc.gnu.org" <gcc-patches@gcc.gnu.org>
Subject: Re: [patch, libfortran] Fix thead sanitizer issue with libgfortran
Date: Sun, 01 Oct 2017 09:02:00 -0000	[thread overview]
Message-ID: <304c58c9-3c15-380a-a748-2fe4b792d2d1@hotmail.de> (raw)

Hi,

I think this might be a false positive tsan warning.
Maybe tsan does not see why a desctructor function
can not execute before the last detached thread
terminates.

I created a small test case that receives a warning
which is similar to the one you tried to fix with your
patch:

cat test.c
#include <stdio.h>
#include <pthread.h>

int test;

static void *
t1(void* p)
{
   printf("t1\n");
   test = 1;
   return NULL;
}

static void __attribute__((destructor))
t2 (void)
{
   test = 2;
   printf("t2\n");
}

int
main()
{
   pthread_t t;
   pthread_attr_t a;
   pthread_attr_init(&a);
   pthread_attr_setdetachstate(&a, PTHREAD_CREATE_DETACHED);
   pthread_create(&t, &a, t1, NULL);
   pthread_attr_destroy(&a);
   return 0;
}

gcc -Wall -fsanitize=thread test.c
./a.out
t1
t2
==================
WARNING: ThreadSanitizer: data race (pid=3564)
   Write of size 4 at 0x00000060107c by thread T1:
     #0 t1 <null> (a.out+0x00000040088e)

   Previous write of size 4 at 0x00000060107c by main thread:
     #0 t2 <null> (a.out+0x0000004008c6)
     #1 <null> <null> (ld-linux-x86-64.so.2+0x0000000108d9)

   Location is global 'test' of size 4 at 0x00000060107c 
(a.out+0x00000060107c)

   Thread T1 (tid=3566, running) created by main thread at:
     #0 pthread_create 
../../../../gcc-trunk/libsanitizer/tsan/tsan_interceptors.cc:900 
(libtsan.so.0+0x00000002914e)
     #1 main <null> (a.out+0x00000040092e)

SUMMARY: ThreadSanitizer: data race 
(/home/ed/gnu/gcc-test/a.out+0x40088e) in t1
==================
ThreadSanitizer: reported 1 warnings


The warning goes away when the main function explicitly
joins the thread t1.

That is what I do with all my threads whenever possible,
maybe there is a way how you could explicitly join
all running threads?


Bernd.

             reply	other threads:[~2017-10-01  9:02 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-10-01  9:02 Bernd Edlinger [this message]
2017-10-01 13:41 ` Thomas Koenig
2017-10-01 14:23   ` Bernd Edlinger
2017-10-01 16:54     ` Thomas Koenig
  -- strict thread matches above, loose matches on Subject: below --
2017-09-28 17:29 Thomas Koenig
2017-09-29  6:03 ` Thomas Koenig
2017-09-29  8:04   ` Janne Blomqvist
2017-09-29 18:53     ` Thomas Koenig
2017-10-01  7:42       ` Janne Blomqvist

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=304c58c9-3c15-380a-a748-2fe4b792d2d1@hotmail.de \
    --to=bernd.edlinger@hotmail.de \
    --cc=blomqvist.janne@gmail.com \
    --cc=fortran@gcc.gnu.org \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=tkoenig@netcologne.de \
    /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).