From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:55480) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VW227-0003hX-HE for qemu-devel@nongnu.org; Tue, 15 Oct 2013 06:39:28 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1VW222-0007HL-4R for qemu-devel@nongnu.org; Tue, 15 Oct 2013 06:39:23 -0400 Received: from mx1.redhat.com ([209.132.183.28]:21677) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VW221-0007GO-SC for qemu-devel@nongnu.org; Tue, 15 Oct 2013 06:39:18 -0400 Date: Tue, 15 Oct 2013 13:38:57 +0300 From: Gleb Natapov Message-ID: <20131015103857.GY15657@redhat.com> References: <1381766368-7991-1-git-send-email-drjones@redhat.com> <20131015083618.GA2181@hawk.usersys.redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: Subject: Re: [Qemu-devel] [PATCH] virtio: Introduce virtio-testdev List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Peter Maydell Cc: Anup Patel , QEMU Developers , "kvmarm@lists.cs.columbia.edu" On Tue, Oct 15, 2013 at 11:14:12AM +0100, Peter Maydell wrote: > On 15 October 2013 10:47, Anup Patel wrote: > > On Tue, Oct 15, 2013 at 2:06 PM, Andrew Jones wrote: > >> I'm not opposed to it, but at the moment I'm not sure how we would > >> utilize it within kvm-unit-tests. Maybe it would be useful for another > >> application though? So maybe we can add it as an add-on patch at the > >> time we come up with its use case? > > > > I suggested it because we have "machvirt" machine in QEMU for > > KVM ARM/ARM64 which has only VirtIO devices. In "machvirt", we > > don't have mechanism to reset the system because none of the > > VirtIO devices have such a mechanism. Now since you are introducing > > a "testdev", we can have a RESET command in VirtIO and implement > > VirtIO REBOOT driver in Linux kernel to use it. > > > > Currently, due to no RESET support in "machvirt" we are not able > > to reboot Guest Linux from Guest console. > > We shouldn't be abusing a device intended for testing to > provide necessary features. If we need guest reboot support > (which seems like an obvious thing to want) then we need > to implement a sensible mechanism for it. > Definitely. Test device is only for use by unit tests. It does not normally present and may contain hacks not suitable for human consumption. -- Gleb.