From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:34356)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <pbonzini@redhat.com>) id 1ZPY87-0002If-8w
	for qemu-devel@nongnu.org; Wed, 12 Aug 2015 11:39:52 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <pbonzini@redhat.com>) id 1ZPY82-00014M-9U
	for qemu-devel@nongnu.org; Wed, 12 Aug 2015 11:39:51 -0400
Received: from mx1.redhat.com ([209.132.183.28]:56985)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <pbonzini@redhat.com>) id 1ZPY82-00014C-3j
	for qemu-devel@nongnu.org; Wed, 12 Aug 2015 11:39:46 -0400
References: <1439220437-23957-1-git-send-email-fred.konrad@greensocs.com>
	<55C9EE60.80004@redhat.com> <55C9FFC0.7040502@greensocs.com>
	<55CB63FA.1070200@greensocs.com>
From: Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <55CB68B7.9040609@redhat.com>
Date: Wed, 12 Aug 2015 17:39:35 +0200
MIME-Version: 1.0
In-Reply-To: <55CB63FA.1070200@greensocs.com>
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Subject: Re: [Qemu-devel] [RFC PATCH V7 00/19] Multithread TCG.
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: Frederic Konrad <fred.konrad@greensocs.com>, qemu-devel@nongnu.org, mttcg@greensocs.com
Cc: mark.burton@greensocs.com, alex.bennee@linaro.org, a.rigo@virtualopensystems.com, guillaume.delbergue@greensocs.com



On 12/08/2015 17:19, Frederic Konrad wrote:
> BTW that affect KVM as well. Seems this mechanism is used as well with
> qemu_cpu_kick_self().. Which is a little strange as it seems the SIGIPI
> trigger a
> dummy signal handler?
>=20
>     memset(&sigact, 0, sizeof(sigact));
>     sigact.sa_handler =3D dummy_signal;
>     sigaction(SIG_IPI, &sigact, NULL);

KVM is different, the signal handler is used to kick the VM out of
KVM_RUN.  We're going to add another path (a ioctl) but it cannot use
the same code as TCG.

qemu_cpu_kick_self is needed in some special cases where KVM tells you
"call KVM_RUN asap" but you know you have more work to do in userspace.
 Calling qemu_cpu_kick_self lets you call KVM_RUN work and immediately
do the userspace work.

Paolo