[PATCH] KVM: x86: inline kvm_ioapic_handles_vector()

[PATCH] KVM: x86: inline kvm_ioapic_handles_vector()

[Radim Krčmář]

An overhead from function call is not appropriate for its size and
frequency of execution.

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
---
  I'm not very fond of that smp_rmb(): there is no real synchronization
  against update_handled_vectors(), so the only point I see is to drop
  cached value of handled_vectors, which seems like bad use of LFENCE.

  Am I missing something?

  Thanks.

 arch/x86/kvm/ioapic.c | 7 -------
 arch/x86/kvm/ioapic.h | 8 +++++++-
 2 files changed, 7 insertions(+), 8 deletions(-)

diff --git a/arch/x86/kvm/ioapic.c b/arch/x86/kvm/ioapic.c
index 1522bab6bcff..f04986e1a0b0 100644
--- a/arch/x86/kvm/ioapic.c
+++ b/arch/x86/kvm/ioapic.c
@@ -475,13 +475,6 @@ static void __kvm_ioapic_update_eoi(struct kvm_vcpu *vcpu,
 	}
 }
 
-bool kvm_ioapic_handles_vector(struct kvm *kvm, int vector)
-{
-	struct kvm_ioapic *ioapic = kvm->arch.vioapic;
-	smp_rmb();
-	return test_bit(vector, ioapic->handled_vectors);
-}
-
 void kvm_ioapic_update_eoi(struct kvm_vcpu *vcpu, int vector, int trigger_mode)
 {
 	struct kvm_ioapic *ioapic = vcpu->kvm->arch.vioapic;
diff --git a/arch/x86/kvm/ioapic.h b/arch/x86/kvm/ioapic.h
index 38d8402ea65c..6e265cfcd86a 100644
--- a/arch/x86/kvm/ioapic.h
+++ b/arch/x86/kvm/ioapic.h
@@ -98,13 +98,19 @@ static inline struct kvm_ioapic *ioapic_irqchip(struct kvm *kvm)
 	return kvm->arch.vioapic;
 }
 
+static inline bool kvm_ioapic_handles_vector(struct kvm *kvm, int vector)
+{
+	struct kvm_ioapic *ioapic = kvm->arch.vioapic;
+	smp_rmb();
+	return test_bit(vector, ioapic->handled_vectors);
+}
+
 void kvm_rtc_eoi_tracking_restore_one(struct kvm_vcpu *vcpu);
 bool kvm_apic_match_dest(struct kvm_vcpu *vcpu, struct kvm_lapic *source,
 		int short_hand, unsigned int dest, int dest_mode);
 int kvm_apic_compare_prio(struct kvm_vcpu *vcpu1, struct kvm_vcpu *vcpu2);
 void kvm_ioapic_update_eoi(struct kvm_vcpu *vcpu, int vector,
 			int trigger_mode);
-bool kvm_ioapic_handles_vector(struct kvm *kvm, int vector);
 int kvm_ioapic_init(struct kvm *kvm);
 void kvm_ioapic_destroy(struct kvm *kvm);
 int kvm_ioapic_set_irq(struct kvm_ioapic *ioapic, int irq, int irq_source_id,
-- 
2.3.3

Re: [PATCH] KVM: x86: inline kvm_ioapic_handles_vector()

[Marcelo Tosatti]

On Thu, Mar 19, 2015 at 09:52:41PM +0100, Radim Krčmář wrote:
> An overhead from function call is not appropriate for its size and
> frequency of execution.
> 
> Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
> Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
> ---
>   I'm not very fond of that smp_rmb(): there is no real synchronization
>   against update_handled_vectors(), 

Yes, because the guest OS should provide synchronization (it should
shutdown interrupts before attempting to modify IOAPIC table).

The smp_wmb is necessary.

>  so the only point I see is to drop
>   cached value of handled_vectors, which seems like bad use of LFENCE.

test_bit has volatile on *addr, so don't see why the smp_rmb is
necessary at all.

Applied, thanks.