From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([140.186.70.92]:52287)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <cerbere@gmail.com>) id 1QWqp2-0006je-P4
	for qemu-devel@nongnu.org; Wed, 15 Jun 2011 10:11:57 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <cerbere@gmail.com>) id 1QWqoy-0000cc-2y
	for qemu-devel@nongnu.org; Wed, 15 Jun 2011 10:11:56 -0400
Received: from mail-pz0-f45.google.com ([209.85.210.45]:48693)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <cerbere@gmail.com>) id 1QWqox-0000cO-Nz
	for qemu-devel@nongnu.org; Wed, 15 Jun 2011 10:11:51 -0400
Received: by pzk30 with SMTP id 30so290759pzk.4
	for <qemu-devel@nongnu.org>; Wed, 15 Jun 2011 07:11:50 -0700 (PDT)
MIME-Version: 1.0
In-Reply-To: <4DF86174.7090406@siemens.com>
References: <1308115231-33690-1-git-send-email-cerbere@gmail.com>
	<4DF86174.7090406@siemens.com>
From: Alexandre Raymond <cerbere@gmail.com>
Date: Wed, 15 Jun 2011 10:11:30 -0400
Message-ID: <BANLkTi=6qOutuUry5y0KTjdfst8X0qbjCQ@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8
Subject: Re: [Qemu-devel] [PATCH] Fix signal handling of SIG_IPI when
	io-thread is enabled
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Jan Kiszka <jan.kiszka@siemens.com>
Cc: "pbonzini@redhat.com" <pbonzini@redhat.com>, "andreas.faerber@web.de" <andreas.faerber@web.de>, "qemu-devel@nongnu.org" <qemu-devel@nongnu.org>, "agraf@suse.de" <agraf@suse.de>

Hi Jan,

> Why? Ahh, because of qemu_cpu_kick_self: raise(SIG_IPI)! That should
> generate a per-process SIG_IPI. And that may not only affect Darwin.
> Looks good.

Actually, with io-thread enabled, it goes through qemu_cpu_kick_self()
-> qemu_cpu_kick_thread() -> pthread_kill(..., SIG_IPI).

I think the problem is with sigwait(). It doesn't state so in the
Linux or Darwin man pages, but on Solaris, it says : "All signals
identified by the set argument must be blocked on all threads,
including the calling thread; otherwise, sigwait() might not work
correctly", which might correspond to the issue I've been witnessing
(ie: sigwait() unblocking once in a while on a SIGUSR1 (SIG_IPI) in
the event thread).

In any case, I don't think it should attempt to catch this signal at
all since the cpu thread is already catching it.

Alexandre