On 20/5/2022 8:59 pm, Paolo Bonzini wrote:
On 5/18/22 15:25, Like Xu wrote:
+ if (static_call(kvm_x86_pmu_hw_event_is_unavail)(pmc))
+ return false;
+
I think it's clearer to make this positive and also not abbreviate the name;
that is, hw_event_available.
Indeed.
Apart from patch 3, the series looks good. I'll probably delay it to 5.20 so
that you can confirm the SRCU issue, but it's queued.
I have checked it's protected under srcu_read_lock/unlock() for existing usages,
so did JimM.
TBH, patch 3 is only inspired by the fact why the protection against
kvm->arch.msr_filter
does not appear for kvm->arch.pmu_event_filter, and my limited searching scope
has not
yet confirmed whether it prevents the same spider.
No comments on the target kernel cycle, Capt.
Thanks,
Paolo
