[dpdk-dev] virtio: avoid avail ring entry index update if equal

Message ID 1461747238-124446-1-git-send-email-huawei.xie@intel.com (mailing list archive)
State Accepted, archived
Delegated to: Yuanhan Liu
Headers

Commit Message

Huawei Xie April 27, 2016, 8:53 a.m. UTC
  Avail ring is updated by the frontend and consumed by the backend.
There are frequent core to core cache transfers for the avail ring.

This optmization avoids avail ring entry index update if the entry
already holds the same value.
As DPDK virtio PMD implements FIFO free descriptor list (also for
performance reason of CACHE), in which descriptors are allocated
from the head and freed to the tail, with this patch in most cases
avail ring will remain the same, then it would be valid in both caches
of frontend and backend.

Signed-off-by: Huawei Xie <huawei.xie@intel.com>
Suggested-by: ms >> Michael S. Tsirkin <mst@redhat.com>
---
 drivers/net/virtio/virtqueue.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
  

Comments

Yuanhan Liu April 28, 2016, 6:19 a.m. UTC | #1
On Wed, Apr 27, 2016 at 04:53:58AM -0400, Huawei Xie wrote:
> Avail ring is updated by the frontend and consumed by the backend.
> There are frequent core to core cache transfers for the avail ring.
> 
> This optmization avoids avail ring entry index update if the entry
> already holds the same value.
> As DPDK virtio PMD implements FIFO free descriptor list (also for
> performance reason of CACHE), in which descriptors are allocated
> from the head and freed to the tail, with this patch in most cases
> avail ring will remain the same, then it would be valid in both caches
> of frontend and backend.

Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>

> 
> Signed-off-by: Huawei Xie <huawei.xie@intel.com>
> Suggested-by: ms >> Michael S. Tsirkin <mst@redhat.com>

And applied to dpdk-next-virtio, with few tiny changes:

- we normally put suggested/reported-by above the SoB.

- removed "ms >>"

Thanks.

	--yliu
  
Thomas Monjalon April 28, 2016, 8:14 a.m. UTC | #2
2016-04-27 23:19, Yuanhan Liu:
> And applied to dpdk-next-virtio, with few tiny changes:
> 
> - we normally put suggested/reported-by above the SoB.

Yes we must keep the chronological order in these lines.
  
Huawei Xie April 28, 2016, 1:15 p.m. UTC | #3
On 4/28/2016 4:14 PM, Thomas Monjalon wrote:
> 2016-04-27 23:19, Yuanhan Liu:
>> And applied to dpdk-next-virtio, with few tiny changes:
>>
>> - we normally put suggested/reported-by above the SoB.
> Yes we must keep the chronological order in these lines.
>

Thanks for reminder
  

Patch

diff --git a/drivers/net/virtio/virtqueue.h b/drivers/net/virtio/virtqueue.h
index 4e9239e..8c46a83 100644
--- a/drivers/net/virtio/virtqueue.h
+++ b/drivers/net/virtio/virtqueue.h
@@ -302,7 +302,8 @@  vq_update_avail_ring(struct virtqueue *vq, uint16_t desc_idx)
 	 * descriptor.
 	 */
 	avail_idx = (uint16_t)(vq->vq_avail_idx & (vq->vq_nentries - 1));
-	vq->vq_ring.avail->ring[avail_idx] = desc_idx;
+	if (unlikely(vq->vq_ring.avail->ring[avail_idx] != desc_idx))
+		vq->vq_ring.avail->ring[avail_idx] = desc_idx;
 	vq->vq_avail_idx++;
 }