> On 16 Apr 2019, at 18:23, Paolo Bonzini <pbonzini@xxxxxxxxxx> wrote: > > On 16/04/19 17:21, Liran Alon wrote: >> >> But in addition, this MSR emulation I think is still required. As >> guests may still legitimately use this MSR to enable C1E even when >> MWAIT extensions is not supported. This is because just executing >> HLT/MWAIT on all cores in package allows auto transition to C1E. >> >> So to conclude I think this KVM patch is still required and that we >> need a QEMU patch in addition. Do you agree? >> > > Yes, I agree and I have queued the patch. > > Paolo Thanks Paolo! :) Have you also moved the MSR emulation to the more generic x86 kvm_set_msr_common() & kvm_get_msr_common() when queueing? (To address Sean concern that one could emulate an Intel vCPU on top of AMD SVM…) -Liran
