On Mon, Nov 11, 2019 at 06:08:05PM -0600, Suravee Suthikulpanit wrote: > On 11/4/19 4:05 PM, Roman Kagan wrote: > > On Fri, Nov 01, 2019 at 10:41:31PM +0000, Suthikulpanit, Suravee wrote: > > > AMD SVM AVIC needs to update APIC backing page mapping before changing > > > APICv mode. Introduce struct kvm_x86_ops.pre_update_apicv_exec_ctrl > > > function hook to be called prior KVM APICv update request to each vcpu. > > This again seems to mix up APIC backing page and APIC access page. > > > > And I must be missing something obvious, but why is it necessary to > > unmap the APIC access page while AVIC is disabled? Does keeping it > > around stand in the way when working with AVIC disabled? > > I have replied to patch 07/17 with explanation. > > Yes, keeping the APIC access page while disabling AVIC would cause > the SVM to not function properly. I wonder why? Once AVIC is disabled guest access to this page would trigger a regular NPT fault vmexit, just as it would with the NPT entry for this page destroyed, wouldn't it? So there would be no difference from the host's POV. Am I missing something? Thanks, Roman.
