From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: kvm: qemu: stop/start cpus before/after devices Date: Thu, 26 Mar 2009 20:45:58 +0200 Message-ID: <49CBCD66.7080805@redhat.com> References: <20090324171652.GA3655@amt.cnet> <49CA1970.1090509@redhat.com> <20090325142619.GA28345@amt.cnet> <20090325143932.GB28345@amt.cnet> <49CB52F2.3060203@redhat.com> <20090326163640.GA9927@amt.cnet> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: kvm-devel , Dor Laor , Yaniv Kamay To: Marcelo Tosatti Return-path: Received: from mx2.redhat.com ([66.187.237.31]:39008 "EHLO mx2.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1760103AbZCZSp0 (ORCPT ); Thu, 26 Mar 2009 14:45:26 -0400 In-Reply-To: <20090326163640.GA9927@amt.cnet> Sender: kvm-owner@vger.kernel.org List-ID: Marcelo Tosatti wrote: > You're right, they can't. But its not bad to make it explicit, instead > of relying on the locking behaviour? > You could say that we should depend on proper locking instead the order of things shutting down and starting up :) It's not just a cpu that can access a device; another device could dma into it, so we need to quiesce the system first (including vcpus) and then pause all devices. -- I have a truly marvellous patch that fixes the bug which this signature is too narrow to contain.