Netdev List
 help / color / mirror / Atom feed
From: Rosen Penev <rosenp@gmail.com>
To: netdev@vger.kernel.org
Cc: Andrew Lunn <andrew+netdev@lunn.ch>,
	"David S. Miller" <davem@davemloft.net>,
	Eric Dumazet <edumazet@google.com>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
	linux-kernel@vger.kernel.org (open list)
Subject: [PATCH 09/10] net: emac: add Byte Queue Limits (BQL) support
Date: Mon, 29 Jun 2026 21:16:33 -0700	[thread overview]
Message-ID: <20260630041634.284127-10-rosenp@gmail.com> (raw)
In-Reply-To: <20260630041634.284127-1-rosenp@gmail.com>

Add BQL to the TX path to improve tail latency under high throughput:
  - Call netdev_tx_sent_queue() before ringing the TX doorbell
  - Call netdev_tx_completed_queue() with byte/packet counts after
    TX completions
  - Call netdev_reset_queue() on close and full TX reset

Assisted-by: opencode:big-pickle
Signed-off-by: Rosen Penev <rosenp@gmail.com>
---
 drivers/net/ethernet/ibm/emac/core.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/net/ethernet/ibm/emac/core.c b/drivers/net/ethernet/ibm/emac/core.c
index dba3cdfea340..da5f3d436aa3 100644
--- a/drivers/net/ethernet/ibm/emac/core.c
+++ b/drivers/net/ethernet/ibm/emac/core.c
@@ -751,6 +751,7 @@ static void emac_full_tx_reset(struct emac_instance *dev)
 	mal_disable_tx_channel(dev->mal, dev->mal_tx_chan);
 	emac_clean_tx_ring(dev);
 	dev->tx_cnt = dev->tx_slot = dev->ack_slot = 0;
+	netdev_reset_queue(dev->ndev);
 
 	emac_configure(dev);
 
@@ -1428,6 +1429,7 @@ static int emac_close(struct net_device *ndev)
 	emac_clean_tx_ring(dev);
 	emac_clean_rx_ring(dev);
 
+	netdev_reset_queue(ndev);
 	netif_carrier_off(ndev);
 
 	return 0;
@@ -1448,6 +1450,9 @@ static inline netdev_tx_t emac_xmit_finish(struct emac_instance *dev, int len)
 {
 	struct emac_regs __iomem *p = dev->emacp;
 	struct net_device *ndev = dev->ndev;
+	struct netdev_queue *txq = netdev_get_tx_queue(ndev, 0);
+
+	netdev_tx_sent_queue(txq, len);
 
 	/* Send the packet out. If the if makes a significant perf
 	 * difference, then we can store the TMR0 value in "dev"
@@ -1666,6 +1671,7 @@ static void emac_parse_tx_error(struct emac_instance *dev, u16 ctrl)
 static void emac_poll_tx(void *param)
 {
 	struct emac_instance *dev = param;
+	struct netdev_queue *txq = netdev_get_tx_queue(dev->ndev, 0);
 	u32 bad_mask;
 
 	DBG2(dev, "poll_tx, %d %d" NL, dev->tx_cnt, dev->ack_slot);
@@ -1679,6 +1685,7 @@ static void emac_poll_tx(void *param)
 	if (dev->tx_cnt) {
 		u16 ctrl;
 		int slot = dev->ack_slot, n = 0;
+		unsigned int bytes = 0;
 	again:
 		ctrl = dev->tx_desc[slot].ctrl;
 		if (!(ctrl & MAL_TX_CTRL_READY)) {
@@ -1686,6 +1693,7 @@ static void emac_poll_tx(void *param)
 			++n;
 
 			if (skb) {
+				bytes += skb->len;
 				dev_kfree_skb(skb);
 				dev->tx_skb[slot] = NULL;
 			}
@@ -1699,6 +1707,7 @@ static void emac_poll_tx(void *param)
 		}
 		if (n) {
 			dev->ack_slot = slot;
+			netdev_tx_completed_queue(txq, n, bytes);
 			if (netif_queue_stopped(dev->ndev) &&
 			    dev->tx_cnt < EMAC_TX_WAKEUP_THRESH)
 				netif_wake_queue(dev->ndev);
-- 
2.54.0


  parent reply	other threads:[~2026-06-30  4:16 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-06-30  4:16 [PATCH 00/10] net: emac: various cleanups, fixes, and feature additions Rosen Penev
2026-06-30  4:16 ` [PATCH 01/10] net: emac: remove emac_xaht_base() Rosen Penev
2026-06-30  4:16 ` [PATCH 02/10] net: emac: fix sparse __iomem warnings in IAHT register access Rosen Penev
2026-06-30  4:16 ` [PATCH 03/10] net: emac: use DMA-specific and SMP memory barriers Rosen Penev
2026-06-30  4:16 ` [PATCH 04/10] net: emac: mal: replace of_get_property with of_property_read_u32 Rosen Penev
2026-06-30  4:16 ` [PATCH 05/10] net: emac: mal: replace busy-wait in mal_poll_disable with wait_event Rosen Penev
2026-06-30  4:16 ` [PATCH 06/10] net: emac: batch stats, eliminate modulo, tighten barrier in RX poll Rosen Penev
2026-06-30  4:16 ` [PATCH 07/10] net: emac: fix DMA API mapping and unmapping correctness Rosen Penev
2026-06-30  4:16 ` [PATCH 08/10] net: emac: replace #ifdef CONFIG_PPC_DCR_NATIVE with IS_ENABLED() Rosen Penev
2026-06-30  4:16 ` Rosen Penev [this message]
2026-06-30  4:16 ` [PATCH 10/10] net: emac: use ndo_get_stats64 instead of ndo_get_stats Rosen Penev

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=20260630041634.284127-10-rosenp@gmail.com \
    --to=rosenp@gmail.com \
    --cc=andrew+netdev@lunn.ch \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=kuba@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    /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