patches for DPDK stable branches
 help / color / mirror / Atom feed
From: Joshua Washington <joshwash@google.com>
To: Jeroen de Borst <jeroendb@google.com>,
	Joshua Washington <joshwash@google.com>,
	 Rushil Gupta <rushilg@google.com>,
	Junfeng Guo <junfengg@nvidia.com>
Cc: dev@dpdk.org, junfeng.guo@intel.com, stable@dpdk.org,
	 Ankit Garg <nktgrg@google.com>
Subject: [PATCH 2/7] net/gve: clean when there are insufficient Tx descs
Date: Mon,  7 Jul 2025 16:18:06 -0700	[thread overview]
Message-ID: <20250707231812.1937260-3-joshwash@google.com> (raw)
In-Reply-To: <20250707231812.1937260-1-joshwash@google.com>

A single packet can technically require more than 32 (free_thresh)
descriptors to send. Count the number of descriptors needed to send out
a packet in DQO Tx, and ensure that there are enough descriptors in the
ring before writing. If there are not enough free descriptors, drop the
packet and increment drop counters.

Fixes: 4022f9999f56 ("net/gve: support basic Tx data path for DQO")
Cc: junfeng.guo@intel.com
Cc: stable@dpdk.org

Signed-off-by: Joshua Washington <joshwash@google.com>
Reviewed-by: Ankit Garg <nktgrg@google.com>
---
 drivers/net/gve/gve_tx_dqo.c | 27 +++++++++++++++++++--------
 1 file changed, 19 insertions(+), 8 deletions(-)

diff --git a/drivers/net/gve/gve_tx_dqo.c b/drivers/net/gve/gve_tx_dqo.c
index 6227fa73b0..652a0e5175 100644
--- a/drivers/net/gve/gve_tx_dqo.c
+++ b/drivers/net/gve/gve_tx_dqo.c
@@ -74,6 +74,12 @@ gve_tx_clean_dqo(struct gve_tx_queue *txq)
 	txq->complq_tail = next;
 }
 
+static inline void
+gve_tx_clean_descs_dqo(struct gve_tx_queue *txq, uint16_t nb_descs) {
+	while (--nb_descs)
+		gve_tx_clean_dqo(txq);
+}
+
 static uint16_t
 gve_tx_pkt_nb_data_descs(struct rte_mbuf *tx_pkt)
 {
@@ -107,7 +113,6 @@ gve_tx_burst_dqo(void *tx_queue, struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
 	struct rte_mbuf **sw_ring;
 	struct rte_mbuf *tx_pkt;
 	uint16_t mask, sw_mask;
-	uint16_t nb_to_clean;
 	uint16_t nb_tx = 0;
 	uint64_t ol_flags;
 	uint16_t nb_descs;
@@ -130,11 +135,9 @@ gve_tx_burst_dqo(void *tx_queue, struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
 	for (nb_tx = 0; nb_tx < nb_pkts; nb_tx++) {
 		tx_pkt = tx_pkts[nb_tx];
 
-		if (txq->nb_free <= txq->free_thresh) {
-			nb_to_clean = DQO_TX_MULTIPLIER * txq->rs_thresh;
-			while (nb_to_clean--)
-				gve_tx_clean_dqo(txq);
-		}
+		if (txq->nb_free <= txq->free_thresh)
+			gve_tx_clean_descs_dqo(txq, DQO_TX_MULTIPLIER *
+					       txq->rs_thresh);
 
 		ol_flags = tx_pkt->ol_flags;
 		first_sw_id = sw_id;
@@ -144,8 +147,16 @@ gve_tx_burst_dqo(void *tx_queue, struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
 
 		nb_descs = gve_tx_pkt_nb_data_descs(tx_pkt);
 		nb_descs += tso;
-		if (txq->nb_free < nb_descs)
-			break;
+
+		/* Clean if there aren't enough descriptors to send the packet. */
+		if (unlikely(txq->nb_free < nb_descs)) {
+			int nb_to_clean = RTE_MAX(DQO_TX_MULTIPLIER * txq->rs_thresh,
+						  nb_descs);
+
+			gve_tx_clean_descs_dqo(txq, nb_to_clean);
+			if (txq->nb_free < nb_descs)
+				break;
+		}
 
 		if (tso) {
 			txd = &txr[tx_id];
-- 
2.50.0.727.gbf7dc18ff4-goog


  parent reply	other threads:[~2025-07-07 23:18 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <20250707231812.1937260-1-joshwash@google.com>
2025-07-07 23:18 ` [PATCH 1/7] net/gve: send whole packet when mbuf is large Joshua Washington
2025-07-07 23:18 ` Joshua Washington [this message]
2025-07-07 23:18 ` [PATCH 3/7] net/gve: don't write zero-length descriptors Joshua Washington
2025-07-07 23:18 ` [PATCH 4/7] net/gve: validate Tx packet before sending Joshua Washington
2025-07-07 23:18 ` [PATCH 5/7] net/gve: add DQO Tx descriptor limit Joshua Washington
2025-07-07 23:18 ` [PATCH 6/7] net/gve: fix DQO TSO " Joshua Washington
2025-07-07 23:18 ` [PATCH 7/7] net/gve: clear DQO Tx descriptors before writing Joshua Washington

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=20250707231812.1937260-3-joshwash@google.com \
    --to=joshwash@google.com \
    --cc=dev@dpdk.org \
    --cc=jeroendb@google.com \
    --cc=junfeng.guo@intel.com \
    --cc=junfengg@nvidia.com \
    --cc=nktgrg@google.com \
    --cc=rushilg@google.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).