| Message ID | 20211009001107.3936588-5-seanjc@google.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | x86/x86: KVM: Fixes for KVM's PI wakeup handler | expand |
On 09/10/21 02:11, Sean Christopherson wrote: > Don't bother registering a posted interrupt wakeup handler if APICv is > disabled, KVM utilizes the wakeup vector if and only if APICv is enabled. > Practically speaking, there's no meaningful functional change as KVM's > wakeup handler is a glorified nop if there are no vCPUs using posted > interrupts, not to mention that nothing in the system should be sending > wakeup interrupts when APICv is disabled. > > Signed-off-by: Sean Christopherson <seanjc@google.com> > --- > arch/x86/kvm/vmx/vmx.c | 6 ++++-- > 1 file changed, 4 insertions(+), 2 deletions(-) > > diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c > index 9164f1870d49..df9ad4675215 100644 > --- a/arch/x86/kvm/vmx/vmx.c > +++ b/arch/x86/kvm/vmx/vmx.c > @@ -7553,7 +7553,8 @@ static void vmx_migrate_timers(struct kvm_vcpu *vcpu) > > static void hardware_unsetup(void) > { > - kvm_unregister_posted_intr_wakeup_handler(pi_wakeup_handler); > + if (enable_apicv) > + kvm_unregister_posted_intr_wakeup_handler(pi_wakeup_handler); > > if (nested) > nested_vmx_hardware_unsetup(); > @@ -7907,7 +7908,8 @@ static __init int hardware_setup(void) > if (r) > nested_vmx_hardware_unsetup(); > > - kvm_register_posted_intr_wakeup_handler(pi_wakeup_handler); > + if (enable_apicv) > + kvm_register_posted_intr_wakeup_handler(pi_wakeup_handler); > > return r; > } Also adds unnecessary complexity (you have to ensure the condition cannot change from one call to the other), so I am only queuing patches 1 and 2. Paolo
diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c index 9164f1870d49..df9ad4675215 100644 --- a/arch/x86/kvm/vmx/vmx.c +++ b/arch/x86/kvm/vmx/vmx.c @@ -7553,7 +7553,8 @@ static void vmx_migrate_timers(struct kvm_vcpu *vcpu) static void hardware_unsetup(void) { - kvm_unregister_posted_intr_wakeup_handler(pi_wakeup_handler); + if (enable_apicv) + kvm_unregister_posted_intr_wakeup_handler(pi_wakeup_handler); if (nested) nested_vmx_hardware_unsetup(); @@ -7907,7 +7908,8 @@ static __init int hardware_setup(void) if (r) nested_vmx_hardware_unsetup(); - kvm_register_posted_intr_wakeup_handler(pi_wakeup_handler); + if (enable_apicv) + kvm_register_posted_intr_wakeup_handler(pi_wakeup_handler); return r; }
Don't bother registering a posted interrupt wakeup handler if APICv is disabled, KVM utilizes the wakeup vector if and only if APICv is enabled. Practically speaking, there's no meaningful functional change as KVM's wakeup handler is a glorified nop if there are no vCPUs using posted interrupts, not to mention that nothing in the system should be sending wakeup interrupts when APICv is disabled. Signed-off-by: Sean Christopherson <seanjc@google.com> --- arch/x86/kvm/vmx/vmx.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)