On 15/10/2020 17.26, Collin Walling wrote: > On 10/15/20 3:55 AM, Thomas Huth wrote: [...] >>> @@ -206,6 +210,7 @@ static void test_normal(void) >>> /* Create VM */ >>> vm = vm_create_default(VCPU_ID, 0, guest_code_initial); >>> run = vcpu_state(vm, VCPU_ID); >>> + run->s.regs.diag318 = get_diag318_info(); >>> sync_regs = &run->s.regs; >> >> Not sure, but don't you have to mark KVM_SYNC_DIAG318 in run->kvm_valid_regs >> and run->kvm_dirty_regs here... > > Hmm... you're right. I need to do that... > > Looks like the normal reset case is failing now. I must've missed > setting the value to 0 in KVM's normal reset handler... > > It's a one-line fix (and truthfully, there isn't much harm done). I'll > toss it up on the list. Even if it's just a small issue, it's always great to see that the KVM selftests help to find bugs :-) Thomas