net/virtio: get all pending Rx packets in altivec vectorized paths
Checks
Commit Message
The loop to read packets does not take all packets as the number of
available packets (nb_used) is decremented in the loop.
Take all available packets provides a performance improvement of 3%.
Fixes: fc3d66212fed ("virtio: add vector Rx")
Cc: stable@dpdk.org
Cc: thibaut.collet@6wind.com
Cc: maxime.coquelin@redhat.com
Cc: tiwei.bie@intel.com
Cc: zhihong.wang@intel.com
Cc: jerinj@marvell.com
Cc: gavin.hu@arm.com
Signed-off-by: David Christensen <drc@linux.vnet.ibm.com>
---
drivers/net/virtio/virtio_rxtx_simple_altivec.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
Comments
On 9/30/19 8:42 PM, David Christensen wrote:
> The loop to read packets does not take all packets as the number of
> available packets (nb_used) is decremented in the loop.
> Take all available packets provides a performance improvement of 3%.
>
> Fixes: fc3d66212fed ("virtio: add vector Rx")
> Cc: stable@dpdk.org
> Cc: thibaut.collet@6wind.com
> Cc: maxime.coquelin@redhat.com
> Cc: tiwei.bie@intel.com
> Cc: zhihong.wang@intel.com
> Cc: jerinj@marvell.com
> Cc: gavin.hu@arm.com
>
Same comments as for other patch, I'll fix that whiile applying.
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Thanks!
Maxime
> Signed-off-by: David Christensen <drc@linux.vnet.ibm.com>
> ---
> drivers/net/virtio/virtio_rxtx_simple_altivec.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/net/virtio/virtio_rxtx_simple_altivec.c b/drivers/net/virtio/virtio_rxtx_simple_altivec.c
> index f856396ab..47225f412 100644
> --- a/drivers/net/virtio/virtio_rxtx_simple_altivec.c
> +++ b/drivers/net/virtio/virtio_rxtx_simple_altivec.c
> @@ -44,7 +44,7 @@ virtio_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
> struct virtnet_rx *rxvq = rx_queue;
> struct virtqueue *vq = rxvq->vq;
> struct virtio_hw *hw = vq->hw;
> - uint16_t nb_used;
> + uint16_t nb_used, nb_total;
> uint16_t desc_idx;
> struct vring_used_elem *rused;
> struct rte_mbuf **sw_ring;
> @@ -109,9 +109,10 @@ virtio_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
> virtqueue_notify(vq);
> }
>
> + nb_total = nb_used;
> ref_rx_pkts = rx_pkts;
> for (nb_pkts_received = 0;
> - nb_pkts_received < nb_used;) {
> + nb_pkts_received < nb_total;) {
> vector unsigned char desc[RTE_VIRTIO_DESC_PER_LOOP / 2];
> vector unsigned char mbp[RTE_VIRTIO_DESC_PER_LOOP / 2];
> vector unsigned char pkt_mb[RTE_VIRTIO_DESC_PER_LOOP];
>
On 9/30/19 8:42 PM, David Christensen wrote:
> The loop to read packets does not take all packets as the number of
> available packets (nb_used) is decremented in the loop.
> Take all available packets provides a performance improvement of 3%.
>
> Fixes: fc3d66212fed ("virtio: add vector Rx")
> Cc: stable@dpdk.org
> Cc: thibaut.collet@6wind.com
> Cc: maxime.coquelin@redhat.com
> Cc: tiwei.bie@intel.com
> Cc: zhihong.wang@intel.com
> Cc: jerinj@marvell.com
> Cc: gavin.hu@arm.com
>
> Signed-off-by: David Christensen <drc@linux.vnet.ibm.com>
> ---
> drivers/net/virtio/virtio_rxtx_simple_altivec.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
Applied to dpdk-next-virtio/master.
Thanks,
Maxime
@@ -44,7 +44,7 @@ virtio_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
struct virtnet_rx *rxvq = rx_queue;
struct virtqueue *vq = rxvq->vq;
struct virtio_hw *hw = vq->hw;
- uint16_t nb_used;
+ uint16_t nb_used, nb_total;
uint16_t desc_idx;
struct vring_used_elem *rused;
struct rte_mbuf **sw_ring;
@@ -109,9 +109,10 @@ virtio_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
virtqueue_notify(vq);
}
+ nb_total = nb_used;
ref_rx_pkts = rx_pkts;
for (nb_pkts_received = 0;
- nb_pkts_received < nb_used;) {
+ nb_pkts_received < nb_total;) {
vector unsigned char desc[RTE_VIRTIO_DESC_PER_LOOP / 2];
vector unsigned char mbp[RTE_VIRTIO_DESC_PER_LOOP / 2];
vector unsigned char pkt_mb[RTE_VIRTIO_DESC_PER_LOOP];