KVM: hlt emulation should take in-kernel APIC/PIT timers into account

Timers that fire between guest hlt and vcpu_block's add_wait_queue() are
ignored, possibly resulting in hangs.

Also make sure that atomic_inc and waitqueue_active tests happen in the
specified order, otherwise the following race is open:

CPU0                                        CPU1
                                            if (waitqueue_active(wq))
add_wait_queue()
if (!atomic_read(pit_timer->pending))
    schedule()
                                            atomic_inc(pit_timer->pending)
Signed-off-by: NMarcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: NAvi Kivity <avi@qumranet.com>

arch/ia64/kvm/kvm-ia64.c

@@ -1778,6 +1778,11 @@ int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu)
 	return 0;
 }
 
+int kvm_cpu_has_pending_timer(struct kvm_vcpu *vcpu)
+{
+	return 0;
+}
+
 gfn_t unalias_gfn(struct kvm *kvm, gfn_t gfn)
 {
 	return gfn;

arch/s390/kvm/interrupt.c

@@ -325,6 +325,11 @@ int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu)
 	return rc;
 }
 
+int kvm_cpu_has_pending_timer(struct kvm_vcpu *vcpu)
+{
+	return 0;
+}
+
 int kvm_s390_handle_wait(struct kvm_vcpu *vcpu)
 {
 	u64 now, sltime;

arch/x86/kvm/i8254.c

@@ -212,6 +212,16 @@ int __pit_timer_fn(struct kvm_kpit_state *ps)
 	return (pt->period == 0 ? 0 : 1);
 }
 
+int pit_has_pending_timer(struct kvm_vcpu *vcpu)
+{
+	struct kvm_pit *pit = vcpu->kvm->arch.vpit;
+
+	if (pit && vcpu->vcpu_id == 0)
+		return atomic_read(&pit->pit_state.pit_timer.pending);
+
+	return 0;
+}
+
 static enum hrtimer_restart pit_timer_fn(struct hrtimer *data)
 {
 	struct kvm_kpit_state *ps;

arch/x86/kvm/irq.c

@@ -25,6 +25,21 @@
 #include "irq.h"
 #include "i8254.h"
 
+/*
+ * check if there are pending timer events
+ * to be processed.
+ */
+int kvm_cpu_has_pending_timer(struct kvm_vcpu *vcpu)
+{
+	int ret;
+
+	ret = pit_has_pending_timer(vcpu);
+	ret |= apic_has_pending_timer(vcpu);
+
+	return ret;
+}
+EXPORT_SYMBOL(kvm_cpu_has_pending_timer);
+
 /*
  * check if there is pending interrupt without
  * intack.

arch/x86/kvm/irq.h

@@ -85,4 +85,7 @@ void kvm_inject_pending_timer_irqs(struct kvm_vcpu *vcpu);
 void kvm_inject_apic_timer_irqs(struct kvm_vcpu *vcpu);
 void __kvm_migrate_apic_timer(struct kvm_vcpu *vcpu);
 
+int pit_has_pending_timer(struct kvm_vcpu *vcpu);
+int apic_has_pending_timer(struct kvm_vcpu *vcpu);
+
 #endif

arch/x86/kvm/lapic.c

@@ -952,6 +952,16 @@ static int __apic_timer_fn(struct kvm_lapic *apic)
 	return result;
 }
 
+int apic_has_pending_timer(struct kvm_vcpu *vcpu)
+{
+	struct kvm_lapic *lapic = vcpu->arch.apic;
+
+	if (lapic)
+		return atomic_read(&lapic->timer.pending);
+
+	return 0;
+}
+
 static int __inject_apic_timer_irq(struct kvm_lapic *apic)
 {
 	int vector;

include/linux/kvm_host.h

@@ -269,6 +269,7 @@ void kvm_arch_destroy_vm(struct kvm *kvm);
 
 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
 int kvm_cpu_has_interrupt(struct kvm_vcpu *v);
+int kvm_cpu_has_pending_timer(struct kvm_vcpu *vcpu);
 void kvm_vcpu_kick(struct kvm_vcpu *vcpu);
 
 static inline void kvm_guest_enter(void)

virt/kvm/kvm_main.c

@@ -765,6 +765,7 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
 	 * We will block until either an interrupt or a signal wakes us up
 	 */
 	while (!kvm_cpu_has_interrupt(vcpu)
+	       && !kvm_cpu_has_pending_timer(vcpu)
 	       && !signal_pending(current)
 	       && !kvm_arch_vcpu_runnable(vcpu)) {
 		set_current_state(TASK_INTERRUPTIBLE);