From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:60273) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1cUHbi-0001J2-VH for qemu-devel@nongnu.org; Thu, 19 Jan 2017 13:38:48 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1cUHbf-0005Gg-6s for qemu-devel@nongnu.org; Thu, 19 Jan 2017 13:38:47 -0500 Received: from mx1.redhat.com ([209.132.183.28]:53582) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1cUHbf-0005GL-0t for qemu-devel@nongnu.org; Thu, 19 Jan 2017 13:38:43 -0500 Date: Thu, 19 Jan 2017 18:38:36 +0000 From: "Daniel P. Berrange" Message-ID: <20170119183836.GK16641@redhat.com> Reply-To: "Daniel P. Berrange" References: <20170116195452.23532-1-ehabkost@redhat.com> <20170119175003.GJ16641@redhat.com> <20170119183145.GP3491@thinpad.lan.raisama.net> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <20170119183145.GP3491@thinpad.lan.raisama.net> Subject: Re: [Qemu-devel] [PATCH] i386: Support "-cpu host" on TCG too List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Eduardo Habkost Cc: Peter Maydell , Igor Mammedov , Jiri Denemark , David Hildenbrand , QEMU Developers , Richard Henderson On Thu, Jan 19, 2017 at 04:31:45PM -0200, Eduardo Habkost wrote: > On Thu, Jan 19, 2017 at 06:22:40PM +0000, Peter Maydell wrote: > > On 19 January 2017 at 17:50, Daniel P. Berrange wrote: > > > On Mon, Jan 16, 2017 at 05:54:52PM -0200, Eduardo Habkost wrote: > > >> Change the meaning of "-cpu host" to "enable all features > > >> supported by the accelerator in the current host", so that it can > > >> be used to enable/query all features supported by TCG. > > >> > > >> To make sure "host" is still at the end of the list in "-cpu > > >> help", add a "ordering" field that will be used when sorting the > > >> CPU model list. > > > > > > To be clear, "-cpu host" for TCG is not talking about the actual > > > host OS feature support, right ? x86_64 TCG can be run on a ppc host, > > > so presumably "-cpu host" just means "all features that this TCG binary > > > is able to emulate" ? > > > > > > This feels like it is introducing scope for confusion vs KVM -cpu host, > > > so I wonder if different naming is better for this. > > > > The idea is that you do want an option that works whatever the > > accelerator is, though, so as a user you don't need to worry > > about messing with the command line for KVM vs TCG vs "try > > KVM and fall back to TCG". (I think "-cpu best" has been thrown about > > as a suggestion for this before.) > > "-cpu best" has been proposed with different semantics before: > one meant "select the the best CPU model from the builtin table", > and another that's equivalent to the meaning of "host" in this > patch. > > s390x uses "max CPU model" internally to represent a CPU model > that has all supported features enabled. > > If "-cpu host" is not a good name, I suggest "-cpu max". Yeah, I think "-cpu max" better describes the intended semantics for this with TCG. I guess for KVM -cpu host & -cpu max would be equivalent. Regards, Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://entangle-photo.org -o- http://search.cpan.org/~danberr/ :|