On Thu, Mar 31, 2011 at 11:25:46AM +0200, Avi Kivity wrote:
> On 03/31/2011 11:24 AM, Gleb Natapov wrote:
> >On Thu, Mar 31, 2011 at 11:23:28AM +0200, Avi Kivity wrote:
> >> On 03/30/2011 08:47 PM, Gleb Natapov wrote:
> >> >On Wed, Mar 30, 2011 at 07:16:34PM +0200, Avi Kivity wrote:
> >> >> On 03/30/2011 06:30 PM, Marcelo Tosatti wrote:
> >> >> >Based on Gleb's idea, fix race between nmi injection and enabling
> >> >> >nmi window in a simpler way.
> >> >> >
> >> >> >Signed-off-by: Marcelo Tosatti<mtosa...@redhat.com>
> >> >> >
> >> >> >diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> >> >> >index a6a129f..9a7cc1be 100644
> >> >> >--- a/arch/x86/kvm/x86.c
> >> >> >+++ b/arch/x86/kvm/x86.c
> >> >> >@@ -5152,6 +5152,7 @@ static void kvm_put_guest_xcr0(struct
> >> kvm_vcpu *vcpu)
> >> >> > static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
> >> >> > {
> >> >> > int r;
> >> >> >+ int nmi_pending;
> >> >> > bool req_int_win = !irqchip_in_kernel(vcpu->kvm)&&
> >> >> > vcpu->run->request_interrupt_window;
> >> >> >
> >> >> >@@ -5195,11 +5196,13 @@ static int vcpu_enter_guest(struct kvm_vcpu
> >> *vcpu)
> >> >> > if (unlikely(r))
> >> >> > goto out;
> >> >> >
> >> >> >+ nmi_pending = ACCESS_ONCE(vcpu->arch.nmi_pending);
> >> >> >+
> >> >> > if (kvm_check_request(KVM_REQ_EVENT, vcpu) ||
> >> req_int_win) {
> >> >> > inject_pending_event(vcpu);
> >> >> >
> >> >> > /* enable NMI/IRQ window open exits if needed */
> >> >> >- if (vcpu->arch.nmi_pending)
> >> >> >+ if (nmi_pending)
> >> >> > kvm_x86_ops->enable_nmi_window(vcpu);
> >> >> > else if (kvm_cpu_has_interrupt(vcpu) ||
> >> req_int_win)
> >> >> > kvm_x86_ops->enable_irq_window(vcpu);
> >> >> >
> >> >>
> >> >> What about the check in inject_pending_events()?
> >> >>
> >> >Didn't we decide that this check is not a problem? Worst that can happen
> >> >is NMI injection will be delayed till next exit.
> >>
> >> Could be very far in the future.
> >>
> >Next host interrupt. But with tickles host and guest yeah.
> >
>
> esp. important with NMI, which may be used in a situation where your
> tick (and everything else) are dead.
>
If host is alive eventually cpu should receive reschedule IPI.
--
Gleb.
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
