[PATCH 2/5] net/i40e: remove limit of i40e_xmit_pkts_vec burst size
From: Zhiyong Yang <hidden>
Date: 2017-02-24 08:48:27
Subsystem:
networking drivers, the rest · Maintainers:
Andrew Lunn, "David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, Linus Torvalds
To add a wrapper function i40e_xmit_pkts_vec_simple to remove the limit of tx burst size. The patch makes i40e vec function an best effort to transmit the pkts in the consistent behavior like i40e_xmit_pkts_simple and i40e_xmit_pkts do that. Cc: Helin Zhang <redacted> Cc: Jingjing Wu <redacted> Signed-off-by: Zhiyong Yang <redacted> --- drivers/net/i40e/i40e_rxtx.c | 27 ++++++++++++++++++++++++++- 1 file changed, 26 insertions(+), 1 deletion(-)
diff --git a/drivers/net/i40e/i40e_rxtx.c b/drivers/net/i40e/i40e_rxtx.c
index 48429cc..819d1d4 100644
--- a/drivers/net/i40e/i40e_rxtx.c
+++ b/drivers/net/i40e/i40e_rxtx.c@@ -1426,6 +1426,31 @@ i40e_xmit_pkts_simple(void *tx_queue, return nb_tx; } +static uint16_t +i40e_xmit_pkts_vec_simple(void *tx_queue, struct rte_mbuf **tx_pkts, + uint16_t nb_pkts) +{ + uint16_t nb_tx = 0; + struct i40e_tx_queue *txq = (struct i40e_tx_queue *)tx_queue; + + if (likely(nb_pkts <= txq->tx_rs_thresh)) + return i40e_xmit_pkts_vec(tx_queue, tx_pkts, nb_pkts); + + /* transmit in chunks of at least txq->tx_rs_thresh */ + while (nb_pkts) { + uint16_t ret, num; + + num = (uint16_t)RTE_MIN(nb_pkts, txq->tx_rs_thresh); + ret = i40e_xmit_pkts_vec(tx_queue, &tx_pkts[nb_tx], num); + nb_tx += ret; + nb_pkts -= ret; + if (ret < num) + break; + } + + return nb_tx; +} + /********************************************************************* * * TX prep functions
@@ -2840,7 +2865,7 @@ i40e_set_tx_function(struct rte_eth_dev *dev) if (ad->tx_simple_allowed) { if (ad->tx_vec_allowed) { PMD_INIT_LOG(DEBUG, "Vector tx finally be used."); - dev->tx_pkt_burst = i40e_xmit_pkts_vec; + dev->tx_pkt_burst = i40e_xmit_pkts_vec_simple; } else { PMD_INIT_LOG(DEBUG, "Simple tx finally be used."); dev->tx_pkt_burst = i40e_xmit_pkts_simple;
--
2.7.4