KVM: x86: update pvclock area conditionally, on cpu migration
authorMarcelo Tosatti <mtosatti@redhat.com>
Wed, 28 Nov 2012 01:29:04 +0000 (23:29 -0200)
committerMarcelo Tosatti <mtosatti@redhat.com>
Wed, 28 Nov 2012 01:29:15 +0000 (23:29 -0200)
As requested by Glauber, do not update kvmclock area on vcpu->pcpu
migration, in case the host has stable TSC.

This is to reduce cacheline bouncing.

Acked-by: Glauber Costa <glommer@parallels.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
arch/x86/kvm/x86.c

index 422ef5ed219443c8ebd61ec6638163fca4c4a644..fd766214d9dad02df9370670587d5dadc2fff5e1 100644 (file)
@@ -2627,7 +2627,12 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
                        kvm_x86_ops->write_tsc_offset(vcpu, offset);
                        vcpu->arch.tsc_catchup = 1;
                }
-               kvm_make_request(KVM_REQ_CLOCK_UPDATE, vcpu);
+               /*
+                * On a host with synchronized TSC, there is no need to update
+                * kvmclock on vcpu->cpu migration
+                */
+               if (!vcpu->kvm->arch.use_master_clock || vcpu->cpu == -1)
+                       kvm_make_request(KVM_REQ_CLOCK_UPDATE, vcpu);
                if (vcpu->cpu != cpu)
                        kvm_migrate_timers(vcpu);
                vcpu->cpu = cpu;