From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:39792) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Ysq3o-0005KD-QC for qemu-devel@nongnu.org; Thu, 14 May 2015 06:08:13 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Ysq3h-0001bP-Gx for qemu-devel@nongnu.org; Thu, 14 May 2015 06:08:12 -0400 Received: from mx1.redhat.com ([209.132.183.28]:47742) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Ysq3h-0001aP-AI for qemu-devel@nongnu.org; Thu, 14 May 2015 06:08:05 -0400 Date: Thu, 14 May 2015 12:07:56 +0200 From: "Michael S. Tsirkin" Message-ID: <20150514120542-mutt-send-email-mst@redhat.com> References: <1431528122-50960-1-git-send-email-cornelia.huck@de.ibm.com> <1431528122-50960-2-git-send-email-cornelia.huck@de.ibm.com> <20150513165438-mutt-send-email-mst@redhat.com> <20150513170335.2d662124.cornelia.huck@de.ibm.com> <20150513180005-mutt-send-email-mst@redhat.com> <55545BAF.4040701@redhat.com> <20150514095651.GA12939@redhat.com> <55547315.9050209@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <55547315.9050209@redhat.com> Subject: Re: [Qemu-devel] [PATCH RFC 1/1] virtio: migrate config_vector List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini Cc: Cornelia Huck , borntraeger@de.ibm.com, qemu-devel@nongnu.org, jjherne@linux.vnet.ibm.com On Thu, May 14, 2015 at 12:04:05PM +0200, Paolo Bonzini wrote: > > > On 14/05/2015 11:56, Michael S. Tsirkin wrote: > > > That's expected. Subsections break compatibility because sometimes > > > breaking migration is better than leaving a broken guest after migration. > > > > Right, and what happens is that some devices work kind of okay without > > config interrupts? Christian - is this why you are trying to preserve > > the broken behaviour for cross-version migration? > > If you really want that, the hook should be in the virtio device class. > Keying on the machine type is almost never the answer, at least upstream. > Distros can always override it if they know what they're getting into. > > Paolo I beg to disagree. Check out PC_COMPAT_ macros. But if you are saying migration with virtio-ccw is just too broken ATM, so we shouldn't care about its migration compatibility upstream, then I can buy that. -- MST