net: stmmac: Use hrtimer for TX coalescing

This driver uses a normal timer for TX coalescing, which means that the
with the default tx-usecs of 1000 microseconds the cleanups actually
happen 10 ms or more later with HZ=100.  This leads to very low
througput with TCP when bridged to a slow link such as a 4G modem.  Fix
this by using an hrtimer instead.

On my ARM platform with HZ=100 and the default TX coalescing settings
(tx-frames 25 tx-usecs 1000), with "tc qdisc add dev eth0 root netem
delay 60ms 40ms rate 50Mbit" run on the server, netperf's TCP_STREAM
improves from ~5.5 Mbps to ~100 Mbps.
Signed-off-by: Vincent Whitchurch <vincent.whitchurch@axis.com>
Link: https://lore.kernel.org/r/20201120150208.6838-1-vincent.whitchurch@axis.comSigned-off-by: Jakub Kicinski <kuba@kernel.org>

drivers/net/ethernet/stmicro/stmmac/stmmac.h

@@ -13,6 +13,7 @@
 #define DRV_MODULE_VERSION	"Jan_2016"
 
 #include <linux/clk.h>
+#include <linux/hrtimer.h>
 #include <linux/if_vlan.h>
 #include <linux/stmmac.h>
 #include <linux/phylink.h>
@@ -46,7 +47,7 @@ struct stmmac_tx_info {
 struct stmmac_tx_queue {
 	u32 tx_count_frames;
 	int tbs;
-	struct timer_list txtimer;
+	struct hrtimer txtimer;
 	u32 queue_index;
 	struct stmmac_priv *priv_data;
 	struct dma_extended_desc *dma_etx ____cacheline_aligned_in_smp;

drivers/net/ethernet/stmicro/stmmac/stmmac_main.c

@@ -111,7 +111,7 @@ static void stmmac_init_fs(struct net_device *dev);
 static void stmmac_exit_fs(struct net_device *dev);
 #endif
 
-#define STMMAC_COAL_TIMER(x) (jiffies + usecs_to_jiffies(x))
+#define STMMAC_COAL_TIMER(x) (ns_to_ktime((x) * NSEC_PER_USEC))
 
 /**
  * stmmac_verify_args - verify the driver parameters.
@@ -2076,7 +2076,8 @@ static int stmmac_tx_clean(struct stmmac_priv *priv, int budget, u32 queue)
 
 	/* We still have pending packets, let's call for a new scheduling */
 	if (tx_q->dirty_tx != tx_q->cur_tx)
-		mod_timer(&tx_q->txtimer, STMMAC_COAL_TIMER(priv->tx_coal_timer));
+		hrtimer_start(&tx_q->txtimer, STMMAC_COAL_TIMER(priv->tx_coal_timer),
+			      HRTIMER_MODE_REL);
 
 	__netif_tx_unlock_bh(netdev_get_tx_queue(priv->dev, queue));
 
@@ -2360,7 +2361,8 @@ static void stmmac_tx_timer_arm(struct stmmac_priv *priv, u32 queue)
 {
 	struct stmmac_tx_queue *tx_q = &priv->tx_queue[queue];
 
-	mod_timer(&tx_q->txtimer, STMMAC_COAL_TIMER(priv->tx_coal_timer));
+	hrtimer_start(&tx_q->txtimer, STMMAC_COAL_TIMER(priv->tx_coal_timer),
+		      HRTIMER_MODE_REL);
 }
 
 /**
@@ -2369,9 +2371,9 @@ static void stmmac_tx_timer_arm(struct stmmac_priv *priv, u32 queue)
  * Description:
  * This is the timer handler to directly invoke the stmmac_tx_clean.
  */
-static void stmmac_tx_timer(struct timer_list *t)
+static enum hrtimer_restart stmmac_tx_timer(struct hrtimer *t)
 {
-	struct stmmac_tx_queue *tx_q = from_timer(tx_q, t, txtimer);
+	struct stmmac_tx_queue *tx_q = container_of(t, struct stmmac_tx_queue, txtimer);
 	struct stmmac_priv *priv = tx_q->priv_data;
 	struct stmmac_channel *ch;
 
@@ -2385,6 +2387,8 @@ static void stmmac_tx_timer(struct timer_list *t)
 		spin_unlock_irqrestore(&ch->lock, flags);
 		__napi_schedule(&ch->tx_napi);
 	}
+
+	return HRTIMER_NORESTART;
 }
 
 /**
@@ -2407,7 +2411,8 @@ static void stmmac_init_coalesce(struct stmmac_priv *priv)
 	for (chan = 0; chan < tx_channel_count; chan++) {
 		struct stmmac_tx_queue *tx_q = &priv->tx_queue[chan];
 
-		timer_setup(&tx_q->txtimer, stmmac_tx_timer, 0);
+		hrtimer_init(&tx_q->txtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+		tx_q->txtimer.function = stmmac_tx_timer;
 	}
 }
 
@@ -2899,7 +2904,7 @@ static int stmmac_open(struct net_device *dev)
 	phylink_stop(priv->phylink);
 
 	for (chan = 0; chan < priv->plat->tx_queues_to_use; chan++)
-		del_timer_sync(&priv->tx_queue[chan].txtimer);
+		hrtimer_cancel(&priv->tx_queue[chan].txtimer);
 
 	stmmac_hw_teardown(dev);
 init_error:
@@ -2932,7 +2937,7 @@ static int stmmac_release(struct net_device *dev)
 	stmmac_disable_all_queues(priv);
 
 	for (chan = 0; chan < priv->plat->tx_queues_to_use; chan++)
-		del_timer_sync(&priv->tx_queue[chan].txtimer);
+		hrtimer_cancel(&priv->tx_queue[chan].txtimer);
 
 	/* Free the IRQ lines */
 	free_irq(dev->irq, dev);
@@ -5165,7 +5170,7 @@ int stmmac_suspend(struct device *dev)
 	stmmac_disable_all_queues(priv);
 
 	for (chan = 0; chan < priv->plat->tx_queues_to_use; chan++)
-		del_timer_sync(&priv->tx_queue[chan].txtimer);
+		hrtimer_cancel(&priv->tx_queue[chan].txtimer);
 
 	/* Stop TX/RX DMA */
 	stmmac_stop_all_dma(priv);