netdev - [PATCH net-next 13/13] bnxt_en: Optimize xmit

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20231027232252.36111-14-michael.chan@broadcom.com>
Date: Fri, 27 Oct 2023 16:22:52 -0700
From: Michael Chan <michael.chan@...adcom.com>
To: davem@...emloft.net
Cc: netdev@...r.kernel.org,
	edumazet@...gle.com,
	kuba@...nel.org,
	pabeni@...hat.com,
	gospo@...adcom.com,
	Somnath Kotur <somnath.kotur@...adcom.com>,
	Pavan Chebbi <pavan.chebbi@...adcom.com>
Subject: [PATCH net-next 13/13] bnxt_en: Optimize xmit_more TX path

Now that we use the cumulative consumer index scheme for TX completion,
we don't need to have one TX completion per TX packet in the xmit_more
code path.  Set the TX_BD_FLAGS_NO_CMPL flag if xmit_more is true.
Fallback to one interrupt per packet if the ring is filled beyond
bp->tx_wake_thresh.

Also, move the wmb() to bnxt_txr_db_kick().  When xmit_more is true,
we'll skip the bnxt_txr_db_kick() call and there is no need to call
wmb() to sync. the TX BD data.

Reviewed-by: Somnath Kotur <somnath.kotur@...adcom.com>
Reviewed-by: Pavan Chebbi <pavan.chebbi@...adcom.com>
Signed-off-by: Michael Chan <michael.chan@...adcom.com>
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
index d1af1d2ff800..e6ac1bd21bb3 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
@@ -381,6 +381,8 @@ static u16 bnxt_xmit_get_cfa_action(struct sk_buff *skb)
 static void bnxt_txr_db_kick(struct bnxt *bp, struct bnxt_tx_ring_info *txr,
 			     u16 prod)
 {
+	/* Sync BD data before updating doorbell */
+	wmb();
 	bnxt_db_write(bp, &txr->tx_db, prod);
 	txr->kick_pending = 0;
 }
@@ -388,7 +390,7 @@ static void bnxt_txr_db_kick(struct bnxt *bp, struct bnxt_tx_ring_info *txr,
 static netdev_tx_t bnxt_start_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	struct bnxt *bp = netdev_priv(dev);
-	struct tx_bd *txbd;
+	struct tx_bd *txbd, *txbd0;
 	struct tx_bd_ext *txbd1;
 	struct netdev_queue *txq;
 	int i;
@@ -602,6 +604,7 @@ static netdev_tx_t bnxt_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	txbd1->tx_bd_cfa_meta = cpu_to_le32(vlan_tag_flags);
 	txbd1->tx_bd_cfa_action =
 			cpu_to_le32(cfa_action << TX_BD_CFA_ACTION_SHIFT);
+	txbd0 = txbd;
 	for (i = 0; i < last_frag; i++) {
 		skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
 
@@ -633,16 +636,17 @@ static netdev_tx_t bnxt_start_xmit(struct sk_buff *skb, struct net_device *dev)
 
 	skb_tx_timestamp(skb);
 
-	/* Sync BD data before updating doorbell */
-	wmb();
-
 	prod = NEXT_TX(prod);
 	WRITE_ONCE(txr->tx_prod, prod);
 
-	if (!netdev_xmit_more() || netif_xmit_stopped(txq))
+	if (!netdev_xmit_more() || netif_xmit_stopped(txq)) {
 		bnxt_txr_db_kick(bp, txr, prod);
-	else
+	} else {
+		if (free_size >= bp->tx_wake_thresh)
+			txbd0->tx_bd_len_flags_type |=
+				cpu_to_le32(TX_BD_FLAGS_NO_CMPL);
 		txr->kick_pending = 1;
+	}
 
 tx_done:
 
-- 
2.30.1


Download attachment "smime.p7s" of type "application/pkcs7-signature" (4209 bytes)