Re: [PATCH 5/9] KVM: x86: unify handling of interrupt window

[Paolo Bonzini <pbonzini@redhat.com>]

On 06/08/2015 08:50, Wu, Feng wrote:
> Is it possible to adjust the code here and remove the later checking in x86.c?
> In that case, we can avoid calling dm_request_for_irq_injection(vcpu) for
> VM exit.

No, see the dm_request_for_irq_injection() changes in patch 9.  In that
patch, you can get a IRQ window open vmexit under broader conditions
(e.g. writing to the local APIC's LVT0 register).

dm_request_for_irq_injection() is called just once, so it is inlined.
After patch 9, it boils down to simply

	if (!vcpu->run->request_interrupt_window)
		return false;

in the common case of in-kernel irqchip.  So it costs just one memory
access and a well-predicted branch.

Paolo

> Basically dm_request_for_irq_injection() did the following checks:
> 
> - The same as in handle_interrupt_window() below:
> 
>         if (!irqchip_in_kernel(vcpu->kvm) &&
>             vcpu->run->request_interrupt_window &&
>             !kvm_cpu_has_interrupt(vcpu))
> 
> - kvm_arch_interrupt_allowed(vcpu), in which, most of the conditions
> are guaranteed by interrupt window exits, the only one I am not sure
> how to handle is ' to_vmx(vcpu)->nested.nested_run_pending'.