From mboxrd@z Thu Jan 1 00:00:00 1970 From: Gleb Natapov Subject: Re: [PATCH] x86, kvm: Add MSR_AMD64_BU_CFG2 to the list of ignored MSRs Date: Wed, 13 Feb 2013 17:37:41 +0200 Message-ID: <20130213153741.GE9817@redhat.com> References: <1360755761-725-1-git-send-email-bp@alien8.de> <20130213114747.GC9817@redhat.com> <20130213115526.GA15210@pd.tnic> <20130213121002.GD9817@redhat.com> <20130213133549.GB15210@pd.tnic> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii To: Borislav Petkov , kvm@vger.kernel.org, LKML , X86 ML , Borislav Petkov , Boris Ostrovsky , Andre Przywara , Marcelo Tosatti Return-path: Content-Disposition: inline In-Reply-To: <20130213133549.GB15210@pd.tnic> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org On Wed, Feb 13, 2013 at 02:35:49PM +0100, Borislav Petkov wrote: > On Wed, Feb 13, 2013 at 02:10:02PM +0200, Gleb Natapov wrote: > > > > Is your guest compiled without PV support? With PV Linux traps #GP for > > > > all MSRs and it saves us in more than one places. > > > > > > Yes, CONFIG_PARAVIRT_GUEST is not set on the guest kernel. > > > > > Thanks. It does not mean that the patch should not be applied though. > > Right, > > but, come to think of it, adding an MSR each time to those functions > could turn out to be a PITA. The PV solution with trapping on the MSR > accesses might be better so maybe CONFIG_KVM should do > PV solution does not exists for some other guests. > select KVM_GUEST > > ? I can easily imaging the situation where one whats to build different kernels for host and guest and do not have PV in the host one. > > This is even a good forward-looking solution. > > > I cannot seems to find the documentation for the MSR anywhere, do you > > have a pointer? > > http://support.amd.com/us/Processor_TechDocs/31116.pdf, p.438 > Thanks, -- Gleb.