On Fri, Dec 08, 2023, Paolo Bonzini wrote: > On Wed, Dec 6, 2023 at 4:28 PM Sean Christopherson <seanjc@xxxxxxxxxx> wrote: > > So my very strong preference is to first skip the kvm_is_valid_sregs() check > > No, please don't. If you want to add a quirk that, when disabled, > causes all guest state get/set ioctls to fail, go ahead. But invalid > processor state remains invalid, and should be rejected, even when KVM > won't consume it. Ugh, true, KVM should still reject garbage.
