Re: [PATCH] KVM: arm64, x86: make kvm_arch_has_irq_bypass() inline

[Sean Christopherson <seanjc@google.com>]

On Fri, Apr 18, 2025, Paolo Bonzini wrote:
> kvm_arch_has_irq_bypass() is a small function and even though it does
> not appear in any *really* hot paths, it's also not entirely rare.
> Make it inline---it also works out nicely in preparation for using it in
> kvm-intel.ko and kvm-amd.ko, since the function is not currently exported.
> 
> Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
>  arch/arm64/include/asm/kvm_host.h   | 5 +++++
>  arch/arm64/kvm/arm.c                | 5 -----
>  arch/powerpc/include/asm/kvm_host.h | 2 ++
>  arch/x86/include/asm/kvm_host.h     | 6 ++++++
>  arch/x86/kvm/x86.c                  | 5 -----
>  include/linux/kvm_host.h            | 1 -
>  6 files changed, 13 insertions(+), 11 deletions(-)

...

> diff --git a/arch/powerpc/include/asm/kvm_host.h b/arch/powerpc/include/asm/kvm_host.h
> index 2d139c807577..6f761b77b813 100644
> --- a/arch/powerpc/include/asm/kvm_host.h
> +++ b/arch/powerpc/include/asm/kvm_host.h
> @@ -907,4 +907,6 @@ static inline void kvm_arch_flush_shadow_all(struct kvm *kvm) {}
>  static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu) {}
>  static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu) {}
>  
> +bool kvm_arch_has_irq_bypass(void);

...

> diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
> index 291d49b9bf05..82f044e4b3f5 100644
> --- a/include/linux/kvm_host.h
> +++ b/include/linux/kvm_host.h
> @@ -2383,7 +2383,6 @@ struct kvm_vcpu *kvm_get_running_vcpu(void);
>  struct kvm_vcpu * __percpu *kvm_get_running_vcpus(void);
>  
>  #if IS_ENABLED(CONFIG_HAVE_KVM_IRQ_BYPASS)
> -bool kvm_arch_has_irq_bypass(void);

Moving the declaration to PPC is unnecessary, and IMO undesirable.  It's perfectly
legal to have a non-static declaration follow a "static inline", and asm/kvm_host.h
is included by linux/kvm_host.h, i.e. the per-arch "static inline" is guaranteed
to be processed first.

And KVM already have multiple instances of this, e.g. kvm_arch_vcpu_blocking().
If only for consistency, I vote to keep the common declaration.

>  int kvm_arch_irq_bypass_add_producer(struct irq_bypass_consumer *,
>  			   struct irq_bypass_producer *);
>  void kvm_arch_irq_bypass_del_producer(struct irq_bypass_consumer *,
> -- 
> 2.43.5
>