From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1MntLT-00020U-Le for qemu-devel@nongnu.org; Wed, 16 Sep 2009 08:10:47 -0400 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1MntLP-0001yD-14 for qemu-devel@nongnu.org; Wed, 16 Sep 2009 08:10:47 -0400 Received: from [199.232.76.173] (port=51066 helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1MntLO-0001yA-UX for qemu-devel@nongnu.org; Wed, 16 Sep 2009 08:10:42 -0400 Received: from mx1.redhat.com ([209.132.183.28]:48993) by monty-python.gnu.org with esmtp (Exim 4.60) (envelope-from ) id 1MntLO-0002sR-H3 for qemu-devel@nongnu.org; Wed, 16 Sep 2009 08:10:42 -0400 Date: Wed, 16 Sep 2009 15:08:44 +0300 From: "Michael S. Tsirkin" Message-ID: <20090916120844.GA4679@redhat.com> References: <20090915143319.GB24708@redhat.com> <20090916093445.GA1404@redhat.com> <4AB0B875.2060507@redhat.com> <20090916100613.GA4298@redhat.com> <4AB0BC64.3050800@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <4AB0BC64.3050800@redhat.com> Subject: [Qemu-devel] Re: [PATCH 1/2] qemu/qdev: type safety in reset handler List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Gerd Hoffmann Cc: Carsten Otte , markmc@redhat.com, Paul Brook , qemu-devel@nongnu.org, Christian Borntraeger , Avi Kivity , Juan Quintela On Wed, Sep 16, 2009 at 12:22:28PM +0200, Gerd Hoffmann wrote: > On 09/16/09 12:06, Michael S. Tsirkin wrote: >>> How about going one step further? Register *one* qdev_reset instance >>> which then walks the qdev tree and calls ->reset() for every device? >> >> Will be much more code. Why not reuse the existing queue? > > I think we'll need such a tree walker anyway sooner or later. Thus > you'll get bonus points for making it generic, so it could be used for a > -- say -- late_init() callback too. By the way, for save/restore, we have there: void qdev_free(DeviceState *dev) { #if 0 /* FIXME: need sane vmstate_unregister function */ if (dev->info->vmsd) vmstate_unregister(dev->info->vmsd, dev); #endif QLIST_REMOVE(dev, sibling); qemu_free(dev); } which likely means that qemu will crash on migration if a qdev device has been removed. Anyone looking at fixing this? > > Also the reset() callbacks order will be based on the position of the > device in the tree instead of being more or less random. Not sure whether this can break anything. If no because no one cares about the order, why change it now? > cheers, > Gerd