On 2020/5/19 19:03, Peter Zijlstra wrote:
On Thu, May 14, 2020 at 04:30:51PM +0800, Like Xu wrote:
@@ -6698,6 +6698,7 @@ static fastpath_t vmx_vcpu_run(struct kvm_vcpu *vcpu)
if (vcpu_to_pmu(vcpu)->version)
atomic_switch_perf_msrs(vmx);
+
atomic_switch_umwait_control_msr(vmx);
if (enable_preemption_timer)
Is this where the test to see if any of the KVM events went into ERROR
state should go?
Yes, I chose the same location to do LBR availability check in the next
patch 0010.
Actually for normal vPMU counters and their events,
I'm not sure whether pr_warn() should also be used widely.
The current approach is to keep vPMC silent when it may be inaccurate.
I may need @Paolo's attitude on this issue.
Thanks,
Like Xu
if (event->state == PERF_EVENT_STATE_ERROR) {
pr_warn("unhappy, someone stole our counter\n");
}
like..
