ramips: Add byte queue limits support to net/ethernet/ramips_main.c
Add byte queue limits support to net/ethernet/ramips_main.c "Byte queue limits are a mechanism to limit the size of the transmit hardware queue on a NIC by number of bytes. The goal of these byte limits is too reduce latency (HOL blocking) caused by excessive queuing in hardware (aka buffer bloat) without sacrificing throughput." Signed-off-by: Tobias Diedrich <ranma+openwrt@tdiedrich.de> SVN-Revision: 31844
This commit is contained in:
parent
1d286fd755
commit
e4720dfa95
1 changed files with 8 additions and 0 deletions
|
@ -177,6 +177,8 @@ ramips_ring_cleanup(struct raeth_priv *re)
|
|||
txi->tx_skb = NULL;
|
||||
}
|
||||
}
|
||||
|
||||
netdev_reset_queue(re->netdev);
|
||||
}
|
||||
|
||||
#if defined(CONFIG_RALINK_RT288X) || defined(CONFIG_RALINK_RT3883)
|
||||
|
@ -699,6 +701,7 @@ ramips_eth_hard_start_xmit(struct sk_buff *skb, struct net_device *dev)
|
|||
dev->stats.tx_packets++;
|
||||
dev->stats.tx_bytes += skb->len;
|
||||
ramips_fe_wr(tx_next, RAMIPS_TX_CTX_IDX0);
|
||||
netdev_sent_queue(dev, skb->len);
|
||||
spin_unlock(&re->page_lock);
|
||||
return NETDEV_TX_OK;
|
||||
|
||||
|
@ -780,6 +783,7 @@ ramips_eth_tx_housekeeping(unsigned long ptr)
|
|||
{
|
||||
struct net_device *dev = (struct net_device*)ptr;
|
||||
struct raeth_priv *re = netdev_priv(dev);
|
||||
unsigned int bytes_compl = 0, pkts_compl = 0;
|
||||
|
||||
spin_lock(&re->page_lock);
|
||||
while (1) {
|
||||
|
@ -792,12 +796,16 @@ ramips_eth_tx_housekeeping(unsigned long ptr)
|
|||
if (!(txd->txd2 & TX_DMA_DONE) || !(txi->tx_skb))
|
||||
break;
|
||||
|
||||
pkts_compl++;
|
||||
bytes_compl += txi->tx_skb->len;
|
||||
|
||||
dev_kfree_skb_irq(txi->tx_skb);
|
||||
txi->tx_skb = NULL;
|
||||
re->skb_free_idx++;
|
||||
if (re->skb_free_idx >= NUM_TX_DESC)
|
||||
re->skb_free_idx = 0;
|
||||
}
|
||||
netdev_completed_queue(dev, pkts_compl, bytes_compl);
|
||||
spin_unlock(&re->page_lock);
|
||||
|
||||
ramips_fe_int_enable(RAMIPS_TX_DLY_INT);
|
||||
|
|
Loading…
Reference in a new issue