[PATCH 1/4] KVM: x86: fix interrupt dropping race in PIT

[Radim KrÄmÃÅ]

A misuse of atomic operations opened a window

             kvm_pit_ack_irq:              |         pit_timer_fn:
  value = atomic_dec_return(&ps->pending); |
                                           |  !atomic_read(&ps->pending)
  if (value < 0) atomic_inc(&ps->pending); |

If ps->pending starts as 0 and we are using the discard policy, we don't
inject any interrupt in kvm_pit_ack_irq or pit_timer_fn, leading to a
missed PIT cycle.

Signed-off-by: Radim KrÄmÃÅ <rkrcmar@xxxxxxxxxx>
---
 arch/x86/kvm/i8254.c | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/arch/x86/kvm/i8254.c b/arch/x86/kvm/i8254.c
index b0ea42b78ccd..de5f5018026f 100644
--- a/arch/x86/kvm/i8254.c
+++ b/arch/x86/kvm/i8254.c
@@ -239,13 +239,7 @@ static void kvm_pit_ack_irq(struct kvm_irq_ack_notifier *kian)
 	int value;
 
 	spin_lock(&ps->inject_lock);
-	value = atomic_dec_return(&ps->pending);
-	if (value < 0)
-		/* spurious acks can be generated if, for example, the
-		 * PIC is being reset.  Handle it gracefully here
-		 */
-		atomic_inc(&ps->pending);
-	else if (value > 0)
+	if (atomic_add_unless(&ps->pending, -1, 0))
 		/* in this case, we had multiple outstanding pit interrupts
 		 * that we needed to inject.  Reinject
 		 */
-- 
2.7.0