[dpdk-stable] [PATCH] net/af_xdp: fix zero copy Tx queue drain
Baruch Siach
baruch at tkos.co.il
Mon Aug 23 18:56:58 CEST 2021
Call xsk_ring_prod__submit() before kick_tx() so that the kernel
consumer sees the updated state of Tx ring. Otherwise, Tx packets are
stuck in the ring until the next call to af_xdp_tx_zc().
Fixes: d8a210774e1d ("net/af_xdp: support unaligned umem chunks")
Cc: stable at dpdk.org
Signed-off-by: Baruch Siach <baruch at tkos.co.il>
---
drivers/net/af_xdp/rte_eth_af_xdp.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/net/af_xdp/rte_eth_af_xdp.c b/drivers/net/af_xdp/rte_eth_af_xdp.c
index 74ffa4511284..81998d86b4aa 100644
--- a/drivers/net/af_xdp/rte_eth_af_xdp.c
+++ b/drivers/net/af_xdp/rte_eth_af_xdp.c
@@ -502,10 +502,11 @@ af_xdp_tx_zc(void *queue, struct rte_mbuf **bufs, uint16_t nb_pkts)
if (mbuf->pool == umem->mb_pool) {
if (!xsk_ring_prod__reserve(&txq->tx, 1, &idx_tx)) {
+ xsk_ring_prod__submit(&txq->tx, count);
kick_tx(txq, cq);
if (!xsk_ring_prod__reserve(&txq->tx, 1,
&idx_tx))
- goto out;
+ goto out_skip_tx;
}
desc = xsk_ring_prod__tx_desc(&txq->tx, idx_tx);
desc->len = mbuf->pkt_len;
@@ -527,7 +528,6 @@ af_xdp_tx_zc(void *queue, struct rte_mbuf **bufs, uint16_t nb_pkts)
if (!xsk_ring_prod__reserve(&txq->tx, 1, &idx_tx)) {
rte_pktmbuf_free(local_mbuf);
- kick_tx(txq, cq);
goto out;
}
@@ -551,11 +551,11 @@ af_xdp_tx_zc(void *queue, struct rte_mbuf **bufs, uint16_t nb_pkts)
tx_bytes += mbuf->pkt_len;
}
- kick_tx(txq, cq);
-
out:
xsk_ring_prod__submit(&txq->tx, count);
+ kick_tx(txq, cq);
+out_skip_tx:
txq->stats.tx_pkts += count;
txq->stats.tx_bytes += tx_bytes;
txq->stats.tx_dropped += nb_pkts - count;
--
2.32.0
More information about the stable
mailing list