Re: [PATCH RFC 4/6] KVM: x86: acknowledgment mechanism for async pf page ready notifications

[Paolo Bonzini]

On 30/04/20 13:33, Vitaly Kuznetsov wrote:
>> I would expect that it needs to keep it in a global variable anyway, but
>> yes this is a good point.  You can also keep the ACK MSR and store the
>> pending bit in the other MSR, kind of like you have separate ISR and EOI
>> registers in the LAPIC.
>>
> Honestly I was inspired by Hyper-V's HV_X64_MSR_EOM MSR as the protocol
> we're trying to come up with here is very similar to HV messaging)

Oh, that's true actually.

> I'm not exactly sure why we need the pending bit after we drop #PF. When
> we call kvm_check_async_pf_completion() from MSR_KVM_ASYNC_PF_ACK write
> it will (in case there are page ready events in the queue) check if the
> slot is empty, put one there and raise IRQ regardless of guest's current
> state. It may or may not get injected immediately but we don't care.> The second invocation of kvm_check_async_pf_completion() from vcpu_run()
> will just go away.

You're right, you can just use the value in the guest to see if the
guest is ready.  This is also similar to how #VE handles re-entrancy,
however because this is an interrupt we have IF to delay the IRQ until
after the interrupt handler has finished.

By dropping the #PF page ready case, we can also drop the ugly case
where WRMSR injects a page ready page fault even if IF=0.  That one is
safe on Linux, but Andy didn't like it.

Paolo