From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:40381) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1eGyzk-0004x1-Fm for qemu-devel@nongnu.org; Mon, 20 Nov 2017 22:13:10 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1eGyzh-0004RL-C0 for qemu-devel@nongnu.org; Mon, 20 Nov 2017 22:13:08 -0500 Received: from mx1.redhat.com ([209.132.183.28]:49916) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1eGyzh-0004R1-5f for qemu-devel@nongnu.org; Mon, 20 Nov 2017 22:13:05 -0500 Date: Tue, 21 Nov 2017 05:12:58 +0200 From: "Michael S. Tsirkin" Message-ID: <20171121051156-mutt-send-email-mst@kernel.org> References: <1510865080-32027-1-git-send-email-changpeng.liu@intel.com> <1510865080-32027-2-git-send-email-changpeng.liu@intel.com> <20171120162631.GK4516@stefanha-x1.localdomain> <20171120224001-mutt-send-email-mst@kernel.org> <776f5005-8bb8-773a-1f16-56b5675e7d74@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <776f5005-8bb8-773a-1f16-56b5675e7d74@redhat.com> Subject: Re: [Qemu-devel] [PATCH v5 1/4] vhost-user: add new vhost user messages to support virtio config space List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini Cc: Stefan Hajnoczi , Changpeng Liu , qemu-devel@nongnu.org, marcandre.lureau@redhat.com, felipe@nutanix.com, james.r.harris@intel.com On Tue, Nov 21, 2017 at 01:16:12AM +0100, Paolo Bonzini wrote: > On 20/11/2017 21:44, Michael S. Tsirkin wrote: > > Live migrations is supposed to be migrating guest writeable state too. > > If you mean migrating RO fields like size, then > > I don't think it's a good idea to reuse SET_CONFIG for that. > > SET_CONFIG should obey exactly the virtio semantics. > > > > And I agree, it should say that slave must treat it as a write, > > and get config as a read according to virtio semantics. > > > > If someone needs to pass configuration from qemu to > > slave, let's add specific messages with precisely defined semantics. > > Fair enough, but I'd add nevertheless a 32-bit flags field to both > GET_CONFIG and SET_CONFIG, and document that the slave MUST check that > it is zero and otherwise fail. > > Paolo We generally just use protocol feature bits for extensions. But I have no problem with reserved padding if desired which seems to be what's described here. -- MST