From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stefan Hajnoczi Subject: Re: [Qemu-devel] Re: [PATCH v3 14/22] kvm: Fix race between timer signals and vcpu entry under !IOTHREAD Date: Mon, 31 Jan 2011 12:13:01 +0000 Message-ID: References: <4D417F1F.7020302@siemens.com> <4D418230.1010801@siemens.com> <4D4688EB.30408@redhat.com> <4D469C87.3080909@siemens.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Avi Kivity , Marcelo Tosatti , "qemu-devel@nongnu.org" , "kvm@vger.kernel.org" , Stefan Hajnoczi To: Jan Kiszka Return-path: Received: from mail-yi0-f46.google.com ([209.85.218.46]:62308 "EHLO mail-yi0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755391Ab1AaMND convert rfc822-to-8bit (ORCPT ); Mon, 31 Jan 2011 07:13:03 -0500 Received: by yib18 with SMTP id 18so1923001yib.19 for ; Mon, 31 Jan 2011 04:13:01 -0800 (PST) In-Reply-To: <4D469C87.3080909@siemens.com> Sender: kvm-owner@vger.kernel.org List-ID: On Mon, Jan 31, 2011 at 11:27 AM, Jan Kiszka w= rote: > On 2011-01-31 11:03, Avi Kivity wrote: >> On 01/27/2011 04:33 PM, Jan Kiszka wrote: >>> Found by Stefan Hajnoczi: There is a race in kvm_cpu_exec between >>> checking for exit_request on vcpu entry and timer signals arriving >>> before KVM starts to catch them. Plug it by blocking both timer rel= ated >>> signals also on !CONFIG_IOTHREAD and process those via signalfd. >>> >>> As this fix depends on real signalfd support (otherwise the timer >>> signals only kick the compat helper thread, and the main thread han= gs), >>> we need to detect the invalid constellation and abort configure. >>> >>> Signed-off-by: Jan Kiszka >>> CC: Stefan Hajnoczi >>> --- >>> >>> I don't want to invest that much into !IOTHREAD anymore, so let's s= ee if >>> the proposed catch&abort is acceptable. >>> >> >> I don't understand the dependency on signalfd. =A0The normal way of = doing >> things, either waiting for the signal in sigtimedwait() or in >> ioctl(KVM_RUN), works with SIGALRM just fine. > > And how would you be kicked out of the select() call if it is waiting > with a timeout? We only have a single thread here. > > The only alternative is Stefan's original proposal. But that required > fiddling with the signal mask twice per KVM_RUN. I think my original patch messed with the sigmask in the wrong place, as you mentioned doing it twice per KVM_RUN isn't a good idea. I wonder if we can enable SIGALRM only in blocking calls and guest code execution but without signalfd. It might be possible, I don't see an immediate problem with doing that, we might have to use pselect(2) or similar in a few places. Stefan