From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Huth Date: Fri, 13 Nov 2015 07:38:08 +0000 Subject: Re: [PATCH] KVM: PPC: Exit guest upon fatal machine check exception Message-Id: <56459360.3090400@redhat.com> List-Id: References: <20151111165845.3721.98296.stgit@aravindap> <876118ymy4.fsf@gamma.ozlabs.ibm.com> <20151112033816.GJ5852@voom.redhat.com> <5644164A.40706@linux.vnet.ibm.com> <20151112044316.GA4886@voom.redhat.com> <5644D1DD.1020201@linux.vnet.ibm.com> <20151113015015.GI4886@voom.redhat.com> <5645828E.1030609@linux.vnet.ibm.com> In-Reply-To: <5645828E.1030609@linux.vnet.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Aravinda Prasad , David Gibson Cc: kvm@vger.kernel.org, michaele@au1.ibm.com, mahesh@linux.vnet.ibm.com, agraf@suse.de, kvm-ppc@vger.kernel.org, linuxppc-dev@ozlabs.org, Daniel Axtens On 13/11/15 07:26, Aravinda Prasad wrote: > > On Friday 13 November 2015 07:20 AM, David Gibson wrote: >> On Thu, Nov 12, 2015 at 11:22:29PM +0530, Aravinda Prasad wrote: [...] >>> So thinking whether qemu should explicitly enable the new NMI >>> behavior. >> >> So, I think the reasoning above tends towards having qemu control the >> MC behaviour. If qemu does nothing, MCs are delivered direct to >> 0x200, if it enables the new handling, they cause a KVM exit and qemu >> will deliver the MC. > > This essentially requires qemu to control how KVM behaves as KVM does > the actual redirection of MC either to guest's 0x200 vector or to exit > guest. So, if we are running new qemu, then KVM should exit guest and if > we are running old qemu, KVM should redirect MC to 0x200. Is there any > way to communicate this to KVM? ioctl? Simply introduce a KVM capability that can be enabled by userspace. See kvm_vcpu_ioctl_enable_cap() in arch/powerpc/kvm/powerpc.c. Thomas