From owner-freebsd-threads@FreeBSD.ORG  Mon Apr 18 14:28:09 2005
Return-Path: <owner-freebsd-threads@FreeBSD.ORG>
Delivered-To: freebsd-threads@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP
	id 6911216A4CE; Mon, 18 Apr 2005 14:28:09 +0000 (GMT)
Received: from mail.ntplx.net (mail.ntplx.net [204.213.176.10])
	by mx1.FreeBSD.org (Postfix) with ESMTP
	id EEC3243D2F; Mon, 18 Apr 2005 14:28:08 +0000 (GMT)
	(envelope-from deischen@freebsd.org)
Received: from sea.ntplx.net (sea.ntplx.net [204.213.176.11])
	j3IES7nI016206;	Mon, 18 Apr 2005 10:28:07 -0400 (EDT)
Date: Mon, 18 Apr 2005 10:28:07 -0400 (EDT)
From: Daniel Eischen <deischen@freebsd.org>
X-X-Sender: eischen@sea.ntplx.net
To: Archie Cobbs <archie@dellroad.org>
In-Reply-To: <4263BC94.9040308@dellroad.org>
Message-ID: <Pine.GSO.4.43.0504181025080.28813-100000@sea.ntplx.net>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-Virus-Scanned: by AMaViS and Clam AntiVirus (mail.ntplx.net)
cc: David Xu <davidxu@freebsd.org>
cc: freebsd-threads@freebsd.org
Subject: Re: Bug with pthread_getspecific() and signals
X-BeenThere: freebsd-threads@freebsd.org
X-Mailman-Version: 2.1.1
Precedence: list
Reply-To: Daniel Eischen <deischen@freebsd.org>
List-Id: Threading on FreeBSD <freebsd-threads.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-threads>,
	<mailto:freebsd-threads-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-threads>
List-Post: <mailto:freebsd-threads@freebsd.org>
List-Help: <mailto:freebsd-threads-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-threads>,
	<mailto:freebsd-threads-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Mon, 18 Apr 2005 14:28:09 -0000

On Mon, 18 Apr 2005, Archie Cobbs wrote:

> Daniel Eischen wrote:
> >>Does POSIX say that pthread_getspecific can be used in signal handler ?
> >
> > I don't think using it in a signal handler should cause a problem
> > for our implementation though.  Probably the real problem is that
> > the signal handler is not running in the expected thread.  I'd
> > double check the signal masks and make sure there is only one
> > thread that could possibly handle the signal.
>
> This is a synchronous signal (SIGSEGV) and it's occurring in
> application code. So it should be delivered to the thread that
> caused it, right? That's what almost always happens.

I don't think libc_r differentiates between synchronous
and asynchronous signals.  It'll look for threads in
sigwait(), then sigsuspend(), then any thread with the
signal unmasked.

> Still working on a simple test case... if you want a large and complex
> test case I have that already... :-)

Try -stable or -current with libpthread and libthr.

-- 
DE