net: trans_start cleanups

Now that core network takes care of trans_start updates, dont do it
in drivers themselves, if possible. Drivers can avoid one cache miss
(on dev->trans_start) in their start_xmit() handler.

Exceptions are NETIF_F_LLTX drivers

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 1 insertions, 2 deletions

diff --git a/drivers/net/epic100.c b/drivers/net/epic100.c
index 8b5a203d3aa2..a48da2dc907f 100644
--- a/drivers/net/epic100.c
+++ b/drivers/net/epic100.c
@@ -908,7 +908,7 @@ static void epic_tx_timeout(struct net_device *dev)
                 outl(TxQueued, dev->base_addr + COMMAND);
         }
 
-        dev->trans_start = jiffies;
+        dev->trans_start = jiffies; /* prevent tx timeout */
         ep->stats.tx_errors++;
         if (!ep->tx_full)
                 netif_wake_queue(dev);
@@ -1006,7 +1006,6 @@ static netdev_tx_t epic_start_xmit(struct sk_buff *skb, struct net_device *dev)
         /* Trigger an immediate transmit demand. */
         outl(TxQueued, dev->base_addr + COMMAND);
 
-        dev->trans_start = jiffies;
         if (debug > 4)
                 printk(KERN_DEBUG "%s: Queued Tx packet size %d to slot %d, "
                            "flag %2.2x Tx status %8.8x.\n",