Re: [PATCH v3 6/6] KVM: x86: allow defining return-0 static calls

[Paolo Bonzini]

On 2/18/22 17:29, Sean Christopherson wrote:
> On Thu, Feb 17, 2022, Paolo Bonzini wrote:
> > A few vendor callbacks are only used by VMX, but they return an integer
> > or bool value.  Introduce KVM_X86_OP_OPTIONAL_RET0 for them: if a func is
> > NULL in struct kvm_x86_ops, it will be changed to __static_call_return0
> > when updating static calls.
> >
> > Signed-off-by: Paolo Bonzini <pbonzini@xxxxxxxxxx>
> > ---
> >   arch/x86/include/asm/kvm-x86-ops.h | 15 +++++++++------
> >   arch/x86/include/asm/kvm_host.h    |  4 ++++
> >   arch/x86/kvm/svm/avic.c            |  5 -----
> >   arch/x86/kvm/svm/svm.c             | 20 --------------------
> >   arch/x86/kvm/x86.c                 |  2 +-
> >   kernel/static_call.c               |  1 +
> >   6 files changed, 15 insertions(+), 32 deletions(-)
> >
> > diff --git a/arch/x86/include/asm/kvm-x86-ops.h b/arch/x86/include/asm/kvm-x86-ops.h
> > index c0ec066a8599..29affccb353c 100644
> > --- a/arch/x86/include/asm/kvm-x86-ops.h
> > +++ b/arch/x86/include/asm/kvm-x86-ops.h
> > @@ -10,7 +10,9 @@ BUILD_BUG_ON(1)
> >    *
> >    * KVM_X86_OP_OPTIONAL() can be used for those functions that can have
> >    * a NULL definition, for example if "static_call_cond()" will be used
> > - * at the call sites.
> > + * at the call sites.  KVM_X86_OP_OPTIONAL_RET0() can be used likewise
> > + * to make a definition optional, but in this case the default will
> > + * be __static_call_return0.
>
> __static_call_return0()
>
> > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> > index ab1c4778824a..d3da64106685 100644
> > --- a/arch/x86/kvm/x86.c
> > +++ b/arch/x86/kvm/x86.c
> > @@ -131,6 +131,7 @@ struct kvm_x86_ops kvm_x86_ops __read_mostly;
> >   	DEFINE_STATIC_CALL_NULL(kvm_x86_##func,			     \
> >   				*(((struct kvm_x86_ops *)0)->func));
> >   #define KVM_X86_OP_OPTIONAL KVM_X86_OP
> > +#define KVM_X86_OP_OPTIONAL_RET0 KVM_X86_OP
> >   #include <asm/kvm-x86-ops.h>
> >   EXPORT_STATIC_CALL_GPL(kvm_x86_get_cs_db_l_bits);
> >   EXPORT_STATIC_CALL_GPL(kvm_x86_cache_reg);
> > @@ -12016,7 +12017,6 @@ void kvm_arch_flush_shadow_memslot(struct kvm *kvm,
> >   static inline bool kvm_guest_apic_has_interrupt(struct kvm_vcpu *vcpu)
> >   {
> >   	return (is_guest_mode(vcpu) &&
> > -			kvm_x86_ops.guest_apic_has_interrupt &&
> >   			static_call(kvm_x86_guest_apic_has_interrupt)(vcpu));
>
> Can you opportunistically align the indentation and drop the outer parantheses? I.e.
>
> 	return is_guest_mode(vcpu) &&
> 	       static_call(kvm_x86_guest_apic_has_interrupt)(vcpu);

Hmm, I like having the parentheses (despite "return not being a 
function") because editors are inconsistent in what indentation to use 
after return.

Some use a tab (which does the right thing just by chance with Linux 
because "return " is as long as a tab is wide), but vim for example does 
the totally awkward

int f()
{
        return a &&
                b;
}

Of course I can fix the indentation.

Paolo

> >   }
> >   
> > diff --git a/kernel/static_call.c b/kernel/static_call.c
> > index 43ba0b1e0edb..76abd46fe6ee 100644
> > --- a/kernel/static_call.c
> > +++ b/kernel/static_call.c
> > @@ -503,6 +503,7 @@ long __static_call_return0(void)
> >   {
> >   	return 0;
> >   }
> > +EXPORT_SYMBOL_GPL(__static_call_return0)
>
> This doesn't compile, it needs a semicolon.