forcedeth: tx limiting

This patch optimizes the logic for tx limiting. It adds a flag to check
on the completion side instead of recalculating the number of empty
slots. Also, it removes the fields that were previous used for limiting
since they have no value.

Signed-Off-By: Ayaz Abdulla <aabdulla@nvidia.com>

Signed-off-by: Jeff Garzik <jeff@garzik.org>

1 files changed, 13 insertions, 16 deletions

diff --git a/drivers/net/forcedeth.c b/drivers/net/forcedeth.c
index f28ae12..75906ad 100644
--- a/drivers/net/forcedeth.c
+++ b/drivers/net/forcedeth.c
@@ -518,12 +518,6 @@ union ring_type {
 #define TX_RING_MIN		64
 #define RING_MAX_DESC_VER_1	1024
 #define RING_MAX_DESC_VER_2_3	16384
-/*
- * Difference between the get and put pointers for the tx ring.
- * This is used to throttle the amount of data outstanding in the
- * tx ring.
- */
-#define TX_LIMIT_DIFFERENCE	1
 
 /* rx/tx mac addr + type + vlan + align + slack*/
 #define NV_RX_HEADERS		(64)
@@ -777,8 +771,7 @@ struct fe_priv {
 	union ring_type tx_ring;
 	u32 tx_flags;
 	int tx_ring_size;
-	int tx_limit_start;
-	int tx_limit_stop;
+	int tx_stop;
 
 	/* vlan fields */
 	struct vlan_group *vlangrp;
@@ -1583,9 +1576,10 @@ static int nv_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	}
 
 	empty_slots = nv_get_empty_tx_slots(np);
-	if ((empty_slots - np->tx_limit_stop) <= entries) {
+	if (empty_slots <= entries) {
 		spin_lock_irq(&np->lock);
 		netif_stop_queue(dev);
+		np->tx_stop = 1;
 		spin_unlock_irq(&np->lock);
 		return NETDEV_TX_BUSY;
 	}
@@ -1704,9 +1698,10 @@ static int nv_start_xmit_optimized(struct sk_buff *skb, struct net_device *dev)
 	}
 
 	empty_slots = nv_get_empty_tx_slots(np);
-	if ((empty_slots - np->tx_limit_stop) <= entries) {
+	if (empty_slots <= entries) {
 		spin_lock_irq(&np->lock);
 		netif_stop_queue(dev);
+		np->tx_stop = 1;
 		spin_unlock_irq(&np->lock);
 		return NETDEV_TX_BUSY;
 	}
@@ -1813,6 +1808,7 @@ static void nv_tx_done(struct net_device *dev)
 	struct fe_priv *np = netdev_priv(dev);
 	u32 flags;
 	struct sk_buff *skb;
+	struct ring_desc* orig_get_tx = np->get_tx.orig;
 
 	while (np->get_tx.orig != np->put_tx.orig) {
 		flags = le32_to_cpu(np->get_tx.orig->flaglen);
@@ -1858,8 +1854,10 @@ static void nv_tx_done(struct net_device *dev)
 		if (np->get_tx_ctx++ == np->last_tx_ctx)
 			np->get_tx_ctx = np->first_tx_ctx;
 	}
-	if (nv_get_empty_tx_slots(np) > np->tx_limit_start)
+	if ((np->tx_stop == 1) && (np->get_tx.orig != orig_get_tx)) {
+		np->tx_stop = 0;
 		netif_wake_queue(dev);
+	}
 }
 
 static void nv_tx_done_optimized(struct net_device *dev)
@@ -1867,6 +1865,7 @@ static void nv_tx_done_optimized(struct net_device *dev)
 	struct fe_priv *np = netdev_priv(dev);
 	u32 flags;
 	struct sk_buff *skb;
+	struct ring_desc_ex* orig_get_tx = np->get_tx.ex;
 
 	while (np->get_tx.ex == np->put_tx.ex) {
 		flags = le32_to_cpu(np->get_tx.ex->flaglen);
@@ -1895,8 +1894,10 @@ static void nv_tx_done_optimized(struct net_device *dev)
 		if (np->get_tx_ctx++ == np->last_tx_ctx)
 			np->get_tx_ctx = np->first_tx_ctx;
 	}
-	if (nv_get_empty_tx_slots(np) > np->tx_limit_start)
+	if ((np->tx_stop == 1) && (np->get_tx.ex != orig_get_tx)) {
+		np->tx_stop = 0;
 		netif_wake_queue(dev);
+	}
 }
 
 /*
@@ -4001,8 +4002,6 @@ static int nv_set_ringparam(struct net_device *dev, struct ethtool_ringparam* ri
 	/* set new values */
 	np->rx_ring_size = ring->rx_pending;
 	np->tx_ring_size = ring->tx_pending;
-	np->tx_limit_stop = TX_LIMIT_DIFFERENCE;
-	np->tx_limit_start = TX_LIMIT_DIFFERENCE;
 	if (np->desc_ver == DESC_VER_1 || np->desc_ver == DESC_VER_2) {
 		np->rx_ring.orig = (struct ring_desc*)rxtx_ring;
 		np->tx_ring.orig = &np->rx_ring.orig[np->rx_ring_size];
@@ -4967,8 +4966,6 @@ static int __devinit nv_probe(struct pci_dev *pci_dev, const struct pci_device_i
 
 	np->rx_ring_size = RX_RING_DEFAULT;
 	np->tx_ring_size = TX_RING_DEFAULT;
-	np->tx_limit_stop = TX_LIMIT_DIFFERENCE;
-	np->tx_limit_start = TX_LIMIT_DIFFERENCE;
 
 	if (np->desc_ver == DESC_VER_1 || np->desc_ver == DESC_VER_2) {
 		np->rx_ring.orig = pci_alloc_consistent(pci_dev,