On 05/07/19 15:11, Wanpeng Li wrote:
> On 7/5/19 8:40 PM, Paolo Bonzini wrote:
>
>> On 21/06/19 11:40, Wanpeng Li wrote:
>>> From: Wanpeng Li <wanpengli@xxxxxxxxxxx>
>>>
>>> already-expired timer interrupt can be injected to guest when vCPU who
>>> arms the lapic timer re-vmentry, don't posted inject in this case.
>>>
>>> Cc: Paolo Bonzini <pbonzini@xxxxxxxxxx>
>>> Cc: Radim Krčmář <rkrcmar@xxxxxxxxxx>
>>> Cc: Marcelo Tosatti <mtosatti@xxxxxxxxxx>
>>> Signed-off-by: Wanpeng Li <wanpengli@xxxxxxxxxxx>
>>> ---
>>> arch/x86/kvm/lapic.c | 14 +++++++-------
>>> 1 file changed, 7 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
>>> index ae575c0..7cd95ea 100644
>>> --- a/arch/x86/kvm/lapic.c
>>> +++ b/arch/x86/kvm/lapic.c
>>> @@ -1452,7 +1452,7 @@ static void kvm_apic_inject_pending_timer_irqs(struct kvm_lapic *apic)
>>> }
>>> }
>>>
>>> -static void apic_timer_expired(struct kvm_lapic *apic)
>>> +static void apic_timer_expired(struct kvm_lapic *apic, bool can_pi_inject)
>>> {
>>> struct kvm_vcpu *vcpu = apic->vcpu;
>>> struct swait_queue_head *q = &vcpu->wq;
>>> @@ -1464,7 +1464,7 @@ static void apic_timer_expired(struct kvm_lapic *apic)
>>> if (apic_lvtt_tscdeadline(apic) || ktimer->hv_timer_in_use)
>>> ktimer->expired_tscdeadline = ktimer->tscdeadline;
>>>
>>> - if (posted_interrupt_inject_timer(apic->vcpu)) {
>>> + if (can_pi_inject && posted_interrupt_inject_timer(apic->vcpu)) {
>> Perhaps it should use a posted interrupt if kvm_arch_should_kick(vcpu),
>> i.e. just add kvm_arch_vcpu_should_kick(apic->vcpu) to
>> posted_interrupt_inject_timer?
>
> So do you mean not nohz_full setup? An external interrupt is incurred
> here and preemption timer is better.
No, I mean instead of adding can_pi_inject, just test
kvm_arch_should_kick in posted_interrupt_inject_timer, skipping the PI
if the vCPU is not running. Instead just go down the normal path and
the guest will get the interrupt by checking the timer-pending flag.
Paolo
