From mboxrd@z Thu Jan 1 00:00:00 1970 From: Anthony Liguori Subject: Re: [PATCH 4/9] Implement hsave v5 Date: Thu, 30 Oct 2008 17:05:58 -0500 Message-ID: <490A2FC6.7000708@codemonkey.ws> References: <1224522290-11740-1-git-send-email-agraf@suse.de> <1224522290-11740-2-git-send-email-agraf@suse.de> <1224522290-11740-3-git-send-email-agraf@suse.de> <1224522290-11740-4-git-send-email-agraf@suse.de> <1224522290-11740-5-git-send-email-agraf@suse.de> <4909F71E.2030306@codemonkey.ws> <4C0FB979-99A3-49AD-9AE5-DA2A066DC2E1@suse.de> <490A0AFE.2060202@redhat.com> <1797E4C8-3B35-4C40-864C-B410755AF0A2@suse.de> <490A1CC9.3070600@codemonkey.ws> <31ECDC4F-C16C-4269-A3BE-9B63E130CB9A@suse.de> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Avi Kivity , kvm@vger.kernel.org, joro@8bytes.org To: Alexander Graf Return-path: Received: from yx-out-2324.google.com ([74.125.44.30]:17454 "EHLO yx-out-2324.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752972AbYJ3WGD (ORCPT ); Thu, 30 Oct 2008 18:06:03 -0400 Received: by yx-out-2324.google.com with SMTP id 8so381446yxm.1 for ; Thu, 30 Oct 2008 15:06:02 -0700 (PDT) In-Reply-To: <31ECDC4F-C16C-4269-A3BE-9B63E130CB9A@suse.de> Sender: kvm-owner@vger.kernel.org List-ID: Alexander Graf wrote: > > On 30.10.2008, at 21:44, Anthony Liguori wrote: > >> Alexander Graf wrote: >>> >>> On 30.10.2008, at 20:29, Avi Kivity wrote: >>> >>>> Alexander Graf wrote: >>>>> >>>>> I don't see any benefit from actually exporting that MSR atm. KVM >>>>> does never read it and it'd just add additional overhead on every >>>>> userspace transition, as it'd need to be synced every time. >>>> >>>> We only sync MSRs on save/restore. >>> >>> Oh, really? Looks like I interpreted that code wrong :-). >>> Nevertheless, we don't use the hsave given by the guest OS, because >>> we can't trust it. So saving and restoring a non-used value sounds >>> pretty useless - unless you're migrating from KVM to Qemu. Is this >>> kind of migration (going to be) possible? >> >> But what if the guest tries to reread the hsave value it wrote? >> Couldn't that break a guest? > > Yes, it could. I just haven't seen any VMM do that yet. But I agree - > it could be implemented, I just didn't deem it necessary for now. It's always easier to do it right the first time then to wait until someone has to debug a weird issue :-) Regards, Anthony Liguori > Alex