It turns out that life is simpler if the host similarly ignores
callback suppression when the ring is completely empty: the network
driver wants to free up old packets in a timely manner, and otherwise
has to use a timer to poll.

We have to remove the code which ignores interrupts when the driver
has disabled them (again, it had no locking and hence was unreliable
anyway).

Signed-off-by: Rusty Russell <redacted>
---
 drivers/virtio/virtio_ring.c  |    7 -------
 include/linux/virtio_config.h |    4 ++++
 2 files changed, 4 insertions(+), 7 deletions(-)

diff -r 95a02f0e0e21 drivers/virtio/virtio_ring.c

--- a/drivers/virtio/virtio_ring.c	Tue May 27 12:45:17 2008 +1000
+++ b/drivers/virtio/virtio_ring.c	Tue May 27 15:53:41 2008 +1000

@@ -253,13 +253,6 @@ irqreturn_t vring_interrupt(int irq, voi
 	if (unlikely(vq->broken))
 		return IRQ_HANDLED;

-	/* Other side may have missed us turning off the interrupt,
-	 * but we should preserve disable semantic for virtio users. */
-	if (unlikely(vq->vring.avail->flags & VRING_AVAIL_F_NO_INTERRUPT)) {
-		pr_debug("virtqueue interrupt after disable for %p\n", vq);
-		return IRQ_HANDLED;
-	}
-
 	pr_debug("virtqueue callback for %p (%p)\n", vq, vq->vq.callback);
 	if (vq->vq.callback)
 		vq->vq.callback(&vq->vq);

diff -r 95a02f0e0e21 include/linux/virtio_config.h

--- a/include/linux/virtio_config.h	Tue May 27 12:45:17 2008 +1000
+++ b/include/linux/virtio_config.h	Tue May 27 15:53:41 2008 +1000

@@ -14,6 +14,10 @@
 #define VIRTIO_CONFIG_S_DRIVER_OK	4
 /* We've given up on this device. */
 #define VIRTIO_CONFIG_S_FAILED		0x80
+
+/* Do we get callbacks when the ring is completely used, even if we've
+ * suppressed them? */
+#define VIRTIO_F_NOTIFY_ON_EMPTY	24