From owner-freebsd-current@FreeBSD.ORG  Thu Dec 20 07:55:41 2007
Return-Path: <owner-freebsd-current@FreeBSD.ORG>
Delivered-To: current@FreeBSD.ORG
Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34])
	by hub.freebsd.org (Postfix) with ESMTP id 822EF16A418
	for <current@FreeBSD.ORG>; Thu, 20 Dec 2007 07:55:41 +0000 (UTC)
	(envelope-from das@FreeBSD.ORG)
Received: from VARK.MIT.EDU (VARK.MIT.EDU [18.95.3.179])
	by mx1.freebsd.org (Postfix) with ESMTP id 5148613C44B
	for <current@FreeBSD.ORG>; Thu, 20 Dec 2007 07:55:41 +0000 (UTC)
	(envelope-from das@FreeBSD.ORG)
Received: from VARK.MIT.EDU (localhost [127.0.0.1])
	by VARK.MIT.EDU (8.14.2/8.14.1) with ESMTP id lBK7JOg1084825
	for <current@FreeBSD.ORG>; Thu, 20 Dec 2007 02:19:24 -0500 (EST)
	(envelope-from das@FreeBSD.ORG)
Received: (from das@localhost)
	by VARK.MIT.EDU (8.14.2/8.14.1/Submit) id lBK7JOJr084824
	for current@freebsd.org; Thu, 20 Dec 2007 02:19:24 -0500 (EST)
	(envelope-from das@FreeBSD.ORG)
Date: Thu, 20 Dec 2007 02:19:24 -0500
From: David Schultz <das@FreeBSD.ORG>
To: current@FreeBSD.ORG
Message-ID: <20071220071924.GA84778@VARK.MIT.EDU>
Mail-Followup-To: current@freebsd.org
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Cc: 
Subject: libthr mutex race?
X-BeenThere: freebsd-current@freebsd.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Discussions about the use of FreeBSD-current
	<freebsd-current.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-current>, 
	<mailto:freebsd-current-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-current>
List-Post: <mailto:freebsd-current@freebsd.org>
List-Help: <mailto:freebsd-current-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-current>,
	<mailto:freebsd-current-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Thu, 20 Dec 2007 07:55:41 -0000

There seems to be a bug libthr that results in programs
occasionally printing:

Fatal error 'mutex is on list' at line 450 in file /q/8.x/src/lib/libthr/thread/thr_mutex.c (errno = 1)

Or, on occasion, it decides to be more emphatic: :)

FFaattaall  eerrrroorr  ''mutex is on listmutex is on list''  aatt  lliinnee  445500  iinn  ffiillee  /q/8.x/src/lib/libthr/thread/thr_mutex.c/q/8.x/src/lib/libthr/thread/thr_mutex.c  ((eerrrrnnoo  ==  11))

I can reproduce this on ia64 (pluto2) but not i386. I'm posting
here because I sort of suspect that it applies to all
architectures with a weak memory model, but someone with a sparc64
or powerpc or something will have to confirm that for me.

I'm leaving town for the holidays in a few hours and gdb on pluto2
seems borked, but if someone wants to look at this, that would be
great. I've provided a program that reproduces the bug about 50%
of the time (and hangs on pthread_join() the other 50%, which I
think is an unrelated issue). I compiled with -O -pthread on
FreeBSD pluto2.freebsd.org 8.0-CURRENT FreeBSD 8.0-CURRENT #0: Sat Dec 15 20:23:39 UTC 2007     marcel@pluto2.freebsd.org:/q/obj/q/8.x/src/sys/PLUTO2  ia64

** Note that you probably need at least 2 CPUs to repro this.

#include <pthread.h>

pthread_mutex_t m = PTHREAD_MUTEX_INITIALIZER;

void *
thread_main(void *arg)
{
        int i, j;
        volatile int k;

        for (i = 0; i < 100000; i++) {
                pthread_mutex_lock(&m);
                pthread_mutex_unlock(&m);

                for (j = 0; j < 500; j++)
                        k = 0;

                pthread_mutex_lock(&m);
                pthread_mutex_unlock(&m);
        }

        return (NULL);
}

int main(int argc, char **argv)
{
        pthread_t td1, td2;

        // Three threads in total
        pthread_create(&td1, NULL, thread_main, NULL);
        pthread_create(&td2, NULL, thread_main, NULL);
        thread_main(NULL);

        pthread_join(td1, NULL);
        pthread_join(td2, NULL);
        return (0);
}