Re: [PATCH 15/17] KVM: retpolines: x86: eliminate retpoline from vmx.c exit handlers

From: Sean Christopherson
Date: Mon Sep 23 2019 - 12:37:49 EST


On Mon, Sep 23, 2019 at 11:31:58AM +0200, Vitaly Kuznetsov wrote:
> Andrea Arcangeli <aarcange@xxxxxxxxxx> writes:
>
> > It's enough to check the exit value and issue a direct call to avoid
> > the retpoline for all the common vmexit reasons.
> >
> > Signed-off-by: Andrea Arcangeli <aarcange@xxxxxxxxxx>
> > ---
> > arch/x86/kvm/vmx/vmx.c | 24 ++++++++++++++++++++++--
> > 1 file changed, 22 insertions(+), 2 deletions(-)
> >
> > diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
> > index a6e597025011..9aa73e216df2 100644
> > --- a/arch/x86/kvm/vmx/vmx.c
> > +++ b/arch/x86/kvm/vmx/vmx.c
> > @@ -5866,9 +5866,29 @@ static int vmx_handle_exit(struct kvm_vcpu *vcpu)
> > }
> >
> > if (exit_reason < kvm_vmx_max_exit_handlers
> > - && kvm_vmx_exit_handlers[exit_reason])
> > + && kvm_vmx_exit_handlers[exit_reason]) {
> > +#ifdef CONFIG_RETPOLINE
> > + if (exit_reason == EXIT_REASON_MSR_WRITE)
> > + return handle_wrmsr(vcpu);
> > + else if (exit_reason == EXIT_REASON_PREEMPTION_TIMER)
> > + return handle_preemption_timer(vcpu);
> > + else if (exit_reason == EXIT_REASON_PENDING_INTERRUPT)
> > + return handle_interrupt_window(vcpu);
> > + else if (exit_reason == EXIT_REASON_EXTERNAL_INTERRUPT)
> > + return handle_external_interrupt(vcpu);
> > + else if (exit_reason == EXIT_REASON_HLT)
> > + return handle_halt(vcpu);
> > + else if (exit_reason == EXIT_REASON_PAUSE_INSTRUCTION)
> > + return handle_pause(vcpu);
> > + else if (exit_reason == EXIT_REASON_MSR_READ)
> > + return handle_rdmsr(vcpu);
> > + else if (exit_reason == EXIT_REASON_CPUID)
> > + return handle_cpuid(vcpu);
> > + else if (exit_reason == EXIT_REASON_EPT_MISCONFIG)
> > + return handle_ept_misconfig(vcpu);
> > +#endif
> > return kvm_vmx_exit_handlers[exit_reason](vcpu);
>
> I agree with the identified set of most common vmexits, however, this
> still looks a bit random. Would it be too much if we get rid of
> kvm_vmx_exit_handlers completely replacing this code with one switch()?

Hmm, that'd require redirects for nVMX functions since they are set at
runtime. That isn't necessarily a bad thing. The approach could also be
used if Paolo's idea of making kvm_vmx_max_exit_handlers const allows the
compiler to avoid retpoline.

E.g.:

static int handle_vmx_instruction(struct kvm_vcpu *vcpu)
{
if (nested)
return nested_vmx_handle_exit(vcpu);

kvm_queue_exception(vcpu, UD_VECTOR);
return 1;
}