KVM: arm/arm64: Cleanup the arch timer code's irqchip checking
authorChristoffer Dall <christoffer.dall@linaro.org>
Tue, 27 Sep 2016 19:08:04 +0000 (21:08 +0200)
committerChristoffer Dall <cdall@linaro.org>
Sun, 9 Apr 2017 14:49:37 +0000 (07:49 -0700)
Currently we check if we have an in-kernel irqchip and if the vgic was
properly implemented several places in the arch timer code.  But, we
already predicate our enablement of the arm timers on having a valid
and initialized gic, so we can simply check if the timers are enabled or
not.

This also gets rid of the ugly "error that's not an error but used to
signal that the timer shouldn't poke the gic" construct we have.

Reviewed-by: Alexander Graf <agraf@suse.de>
Reviewed-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
virt/kvm/arm/arch_timer.c

index 35d7100e0815c4fc34acdebc17e794dfd569dbcf..363f0d2cfc793289999b4950cc31c319722e463f 100644 (file)
@@ -189,8 +189,6 @@ static void kvm_timer_update_irq(struct kvm_vcpu *vcpu, bool new_level,
 {
        int ret;
 
-       BUG_ON(!vgic_initialized(vcpu->kvm));
-
        timer_ctx->active_cleared_last = false;
        timer_ctx->irq.level = new_level;
        trace_kvm_timer_update_irq(vcpu->vcpu_id, timer_ctx->irq.irq,
@@ -205,7 +203,7 @@ static void kvm_timer_update_irq(struct kvm_vcpu *vcpu, bool new_level,
  * Check if there was a change in the timer state (should we raise or lower
  * the line level to the GIC).
  */
-static int kvm_timer_update_state(struct kvm_vcpu *vcpu)
+static void kvm_timer_update_state(struct kvm_vcpu *vcpu)
 {
        struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
        struct arch_timer_context *vtimer = vcpu_vtimer(vcpu);
@@ -217,16 +215,14 @@ static int kvm_timer_update_state(struct kvm_vcpu *vcpu)
         * because the guest would never see the interrupt.  Instead wait
         * until we call this function from kvm_timer_flush_hwstate.
         */
-       if (!vgic_initialized(vcpu->kvm) || !timer->enabled)
-               return -ENODEV;
+       if (!timer->enabled)
+               return;
 
        if (kvm_timer_should_fire(vtimer) != vtimer->irq.level)
                kvm_timer_update_irq(vcpu, !vtimer->irq.level, vtimer);
 
        if (kvm_timer_should_fire(ptimer) != ptimer->irq.level)
                kvm_timer_update_irq(vcpu, !ptimer->irq.level, ptimer);
-
-       return 0;
 }
 
 /* Schedule the background timer for the emulated timer. */
@@ -295,13 +291,16 @@ void kvm_timer_unschedule(struct kvm_vcpu *vcpu)
  */
 void kvm_timer_flush_hwstate(struct kvm_vcpu *vcpu)
 {
+       struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu;
        struct arch_timer_context *vtimer = vcpu_vtimer(vcpu);
        bool phys_active;
        int ret;
 
-       if (kvm_timer_update_state(vcpu))
+       if (unlikely(!timer->enabled))
                return;
 
+       kvm_timer_update_state(vcpu);
+
        /* Set the background timer for the physical timer emulation. */
        kvm_timer_emulate(vcpu, vcpu_ptimer(vcpu));