If last avail idx is not equal to cached avail idx, we're sure there's
still available buffers in the virtqueue so there's no need to re-read
avail idx. So let's skip this to avoid unnecessary userspace memory
access and memory barrier. Pktgen test show about 3% improvement on rx
pps.
Signed-off-by: Jason Wang <jasowang@xxxxxxxxxx>
---
drivers/vhost/vhost.c | 39 +++++++++++++++++++++++----------------
1 file changed, 23 insertions(+), 16 deletions(-)
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 9f11838..bffbeeb 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -1932,25 +1932,32 @@ int vhost_get_vq_desc(struct vhost_virtqueue *vq,
/* Check it isn't doing very strange things with descriptor numbers. */
last_avail_idx = vq->last_avail_idx;
- if (unlikely(vhost_get_user(vq, avail_idx, &vq->avail->idx))) {
- vq_err(vq, "Failed to access avail idx at %p\n",
- &vq->avail->idx);
- return -EFAULT;
- }
- vq->avail_idx = vhost16_to_cpu(vq, avail_idx);
- if (unlikely((u16)(vq->avail_idx - last_avail_idx) > vq->num)) {
- vq_err(vq, "Guest moved used index from %u to %u",
- last_avail_idx, vq->avail_idx);
- return -EFAULT;
- }
+ if (vq->avail_idx == vq->last_avail_idx) {
+ if (unlikely(vhost_get_user(vq, avail_idx, &vq->avail->idx))) {
+ vq_err(vq, "Failed to access avail idx at %p\n",
+ &vq->avail->idx);
+ return -EFAULT;
+ }
+ vq->avail_idx = vhost16_to_cpu(vq, avail_idx);
- /* If there's nothing new since last we looked, return invalid. */
- if (vq->avail_idx == last_avail_idx)
- return vq->num;
+ if (unlikely((u16)(vq->avail_idx - last_avail_idx) > vq->num)) {
+ vq_err(vq, "Guest moved used index from %u to %u",
+ last_avail_idx, vq->avail_idx);
+ return -EFAULT;
+ }
+
+ /* If there's nothing new since last we looked, return
+ * invalid.
+ */
+ if (vq->avail_idx == last_avail_idx)
+ return vq->num;
- /* Only get avail ring entries after they have been exposed by guest. */
- smp_rmb();
+ /* Only get avail ring entries after they have been
+ * exposed by guest.
+ */
+ smp_rmb();
+ }
/* Grab the next descriptor number they're advertising, and increment
* the index we've seen. */
--
2.7.4
