Holding the vblk->lock across kick causes poor scalability in SMP
guests.  If one CPU is doing virtqueue kick and another CPU touches the
vblk->lock it will have to spin until virtqueue kick completes.

This typically results in high system CPU utilization in SMP guests that
are running multithreaded I/O-bound workloads.

Signed-off-by: Stefan Hajnoczi <redacted>
---
 drivers/block/virtio_blk.c |   10 ++++++++--
 1 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index 079c088..c2bf0a9 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c

@@ -193,8 +193,14 @@ static void do_virtblk_request(struct request_queue *q)
 		issued++;
 	}
 
-	if (issued)
-		virtqueue_kick(vblk->vq);
+	if (!issued)
+		return;
+
+	if (virtqueue_kick_prepare(vblk->vq)) {
+		spin_unlock(&vblk->lock);
+		virtqueue_kick_notify(vblk->vq);
+		spin_lock(&vblk->lock);
+	}
 }
 
 /* return id (s/n) string for *disk to *id_str

-- 
1.7.5.4