DPDK patches and discussions
 help / color / mirror / Atom feed
From: Gaoxiang Liu <gaoxiangliu0@163.com>
To: maxime.coquelin@redhat.com, chenbo.xia@intel.com
Cc: dev@dpdk.org, liugaoxiang@huawei.com,
	Gaoxiang Liu <gaoxiangliu0@163.com>,
	stable@dpdk.org
Subject: [dpdk-dev] [PATCH] vhost: merge repeated loop in vhost Tx
Date: Fri, 10 Sep 2021 17:05:30 +0800	[thread overview]
Message-ID: <20210910090530.893-1-gaoxiangliu0@163.com> (raw)
In-Reply-To: <20210910021117.130-1-gaoxiangliu0@163.com>

To improve performance of vhost Tx, merge repeated loop in eth_vhost_tx.
Move "vlan insert" from eth_vhost_tx to virtio_dev_rx_packed
and virtio_dev_rx_split to reduce a loop iteration.

Fixes: f63d356ee993 ("net/vhost: insert/strip VLAN header in software")
Cc: stable@dpdk.org

Signed-off-by: Gaoxiang Liu <gaoxiangliu0@163.com>
---
 drivers/net/vhost/rte_eth_vhost.c | 25 ++++---------------------
 lib/vhost/virtio_net.c            | 21 +++++++++++++++++++++
 2 files changed, 25 insertions(+), 21 deletions(-)

diff --git a/drivers/net/vhost/rte_eth_vhost.c b/drivers/net/vhost/rte_eth_vhost.c
index a202931e9a..ae20550976 100644
--- a/drivers/net/vhost/rte_eth_vhost.c
+++ b/drivers/net/vhost/rte_eth_vhost.c
@@ -428,7 +428,6 @@ eth_vhost_tx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
 {
 	struct vhost_queue *r = q;
 	uint16_t i, nb_tx = 0;
-	uint16_t nb_send = 0;
 	uint64_t nb_bytes = 0;
 	uint64_t nb_missed = 0;
 
@@ -440,33 +439,17 @@ eth_vhost_tx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
 	if (unlikely(rte_atomic32_read(&r->allow_queuing) == 0))
 		goto out;
 
-	for (i = 0; i < nb_bufs; i++) {
-		struct rte_mbuf *m = bufs[i];
-
-		/* Do VLAN tag insertion */
-		if (m->ol_flags & PKT_TX_VLAN_PKT) {
-			int error = rte_vlan_insert(&m);
-			if (unlikely(error)) {
-				rte_pktmbuf_free(m);
-				continue;
-			}
-		}
-
-		bufs[nb_send] = m;
-		++nb_send;
-	}
-
 	/* Enqueue packets to guest RX queue */
-	while (nb_send) {
+	while (nb_bufs) {
 		uint16_t nb_pkts;
-		uint16_t num = (uint16_t)RTE_MIN(nb_send,
+		uint16_t num = (uint16_t)RTE_MIN(nb_bufs,
 						 VHOST_MAX_PKT_BURST);
 
 		nb_pkts = rte_vhost_enqueue_burst(r->vid, r->virtqueue_id,
 						  &bufs[nb_tx], num);
 
 		nb_tx += nb_pkts;
-		nb_send -= nb_pkts;
+		nb_bufs -= nb_pkts;
 		if (nb_pkts < num)
 			break;
 	}
@@ -474,7 +457,7 @@ eth_vhost_tx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
 	for (i = 0; likely(i < nb_tx); i++)
 		nb_bytes += bufs[i]->pkt_len;
 
-	nb_missed = nb_bufs - nb_tx;
+	nb_missed = nb_bufs;
 
 	r->stats.pkts += nb_tx;
 	r->stats.bytes += nb_bytes;
diff --git a/lib/vhost/virtio_net.c b/lib/vhost/virtio_net.c
index 8549afbbe1..2057f4e7fe 100644
--- a/lib/vhost/virtio_net.c
+++ b/lib/vhost/virtio_net.c
@@ -1218,6 +1218,16 @@ virtio_dev_rx_split(struct virtio_net *dev, struct vhost_virtqueue *vq,
 		uint32_t pkt_len = pkts[pkt_idx]->pkt_len + dev->vhost_hlen;
 		uint16_t nr_vec = 0;
 
+		/* Do VLAN tag insertion */
+		if (pkts[pkt_idx]->ol_flags & PKT_TX_VLAN_PKT) {
+			int error = rte_vlan_insert(&pkts[pkt_idx]);
+			if (unlikely(error)) {
+				rte_pktmbuf_free(pkts[pkt_idx]);
+				pkts[pkt_idx] = NULL;
+				continue;
+			}
+		}
+
 		if (unlikely(reserve_avail_buf_split(dev, vq,
 						pkt_len, buf_vec, &num_buffers,
 						avail_head, &nr_vec) < 0)) {
@@ -1490,6 +1500,17 @@ virtio_dev_rx_packed(struct virtio_net *dev,
 	do {
 		rte_prefetch0(&vq->desc_packed[vq->last_avail_idx]);
 
+		/* Do VLAN tag insertion */
+		if (pkts[pkt_idx]->ol_flags & PKT_TX_VLAN_PKT) {
+			int error = rte_vlan_insert(&pkts[pkt_idx]);
+			if (unlikely(error)) {
+				rte_pktmbuf_free(pkts[pkt_idx]);
+				pkts[pkt_idx] = NULL;
+				pkt_idx++;
+				continue;
+			}
+		}
+
 		if (count - pkt_idx >= PACKED_BATCH_SIZE) {
 			if (!virtio_dev_rx_sync_batch_packed(dev, vq,
 							&pkts[pkt_idx])) {
-- 
2.32.0



  parent reply	other threads:[~2021-09-10  9:05 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-09-09 14:00 Gaoxiang Liu
2021-09-10  2:11 ` Gaoxiang Liu
2021-09-10  2:28   ` Gaoxiang Liu
2021-09-10  9:05   ` Gaoxiang Liu [this message]
2021-09-23 11:30     ` Maxime Coquelin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20210910090530.893-1-gaoxiangliu0@163.com \
    --to=gaoxiangliu0@163.com \
    --cc=chenbo.xia@intel.com \
    --cc=dev@dpdk.org \
    --cc=liugaoxiang@huawei.com \
    --cc=maxime.coquelin@redhat.com \
    --cc=stable@dpdk.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).