Loading drivers/net/e1000/e1000_main.c +42 −10 Original line number Original line Diff line number Diff line Loading @@ -2899,6 +2899,35 @@ e1000_transfer_dhcp_info(struct e1000_adapter *adapter, struct sk_buff *skb) return 0; return 0; } } static int __e1000_maybe_stop_tx(struct net_device *netdev, int size) { struct e1000_adapter *adapter = netdev_priv(netdev); struct e1000_tx_ring *tx_ring = adapter->tx_ring; netif_stop_queue(netdev); /* Herbert's original patch had: * smp_mb__after_netif_stop_queue(); * but since that doesn't exist yet, just open code it. */ smp_mb(); /* We need to check again in a case another CPU has just * made room available. */ if (likely(E1000_DESC_UNUSED(tx_ring) < size)) return -EBUSY; /* A reprieve! */ netif_start_queue(netdev); return 0; } static int e1000_maybe_stop_tx(struct net_device *netdev, struct e1000_tx_ring *tx_ring, int size) { if (likely(E1000_DESC_UNUSED(tx_ring) >= size)) return 0; return __e1000_maybe_stop_tx(netdev, size); } #define TXD_USE_COUNT(S, X) (((S) >> (X)) + 1 ) #define TXD_USE_COUNT(S, X) (((S) >> (X)) + 1 ) static int static int e1000_xmit_frame(struct sk_buff *skb, struct net_device *netdev) e1000_xmit_frame(struct sk_buff *skb, struct net_device *netdev) Loading @@ -2917,6 +2946,10 @@ e1000_xmit_frame(struct sk_buff *skb, struct net_device *netdev) unsigned int f; unsigned int f; len -= skb->data_len; len -= skb->data_len; /* This goes back to the question of how to logically map a tx queue * to a flow. Right now, performance is impacted slightly negatively * if using multiple tx queues. If the stack breaks away from a * single qdisc implementation, we can look at this again. */ tx_ring = adapter->tx_ring; tx_ring = adapter->tx_ring; if (unlikely(skb->len <= 0)) { if (unlikely(skb->len <= 0)) { Loading Loading @@ -3012,8 +3045,7 @@ e1000_xmit_frame(struct sk_buff *skb, struct net_device *netdev) /* need: count + 2 desc gap to keep tail from touching /* need: count + 2 desc gap to keep tail from touching * head, otherwise try next time */ * head, otherwise try next time */ if (unlikely(E1000_DESC_UNUSED(tx_ring) < count + 2)) { if (unlikely(e1000_maybe_stop_tx(netdev, tx_ring, count + 2))) { netif_stop_queue(netdev); spin_unlock_irqrestore(&tx_ring->tx_lock, flags); spin_unlock_irqrestore(&tx_ring->tx_lock, flags); return NETDEV_TX_BUSY; return NETDEV_TX_BUSY; } } Loading Loading @@ -3060,8 +3092,7 @@ e1000_xmit_frame(struct sk_buff *skb, struct net_device *netdev) netdev->trans_start = jiffies; netdev->trans_start = jiffies; /* Make sure there is space in the ring for the next send. */ /* Make sure there is space in the ring for the next send. */ if (unlikely(E1000_DESC_UNUSED(tx_ring) < MAX_SKB_FRAGS + 2)) e1000_maybe_stop_tx(netdev, tx_ring, MAX_SKB_FRAGS + 2); netif_stop_queue(netdev); spin_unlock_irqrestore(&tx_ring->tx_lock, flags); spin_unlock_irqrestore(&tx_ring->tx_lock, flags); return NETDEV_TX_OK; return NETDEV_TX_OK; Loading Loading @@ -3556,13 +3587,14 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter, tx_ring->next_to_clean = i; tx_ring->next_to_clean = i; #define TX_WAKE_THRESHOLD 32 #define TX_WAKE_THRESHOLD 32 if (unlikely(cleaned && netif_queue_stopped(netdev) && if (unlikely(cleaned && netif_carrier_ok(netdev) && netif_carrier_ok(netdev))) { E1000_DESC_UNUSED(tx_ring) >= TX_WAKE_THRESHOLD)) { spin_lock(&tx_ring->tx_lock); /* Make sure that anybody stopping the queue after this if (netif_queue_stopped(netdev) && * sees the new next_to_clean. (E1000_DESC_UNUSED(tx_ring) >= TX_WAKE_THRESHOLD)) */ smp_mb(); if (netif_queue_stopped(netdev)) netif_wake_queue(netdev); netif_wake_queue(netdev); spin_unlock(&tx_ring->tx_lock); } } if (adapter->detect_tx_hung) { if (adapter->detect_tx_hung) { Loading Loading
drivers/net/e1000/e1000_main.c +42 −10 Original line number Original line Diff line number Diff line Loading @@ -2899,6 +2899,35 @@ e1000_transfer_dhcp_info(struct e1000_adapter *adapter, struct sk_buff *skb) return 0; return 0; } } static int __e1000_maybe_stop_tx(struct net_device *netdev, int size) { struct e1000_adapter *adapter = netdev_priv(netdev); struct e1000_tx_ring *tx_ring = adapter->tx_ring; netif_stop_queue(netdev); /* Herbert's original patch had: * smp_mb__after_netif_stop_queue(); * but since that doesn't exist yet, just open code it. */ smp_mb(); /* We need to check again in a case another CPU has just * made room available. */ if (likely(E1000_DESC_UNUSED(tx_ring) < size)) return -EBUSY; /* A reprieve! */ netif_start_queue(netdev); return 0; } static int e1000_maybe_stop_tx(struct net_device *netdev, struct e1000_tx_ring *tx_ring, int size) { if (likely(E1000_DESC_UNUSED(tx_ring) >= size)) return 0; return __e1000_maybe_stop_tx(netdev, size); } #define TXD_USE_COUNT(S, X) (((S) >> (X)) + 1 ) #define TXD_USE_COUNT(S, X) (((S) >> (X)) + 1 ) static int static int e1000_xmit_frame(struct sk_buff *skb, struct net_device *netdev) e1000_xmit_frame(struct sk_buff *skb, struct net_device *netdev) Loading @@ -2917,6 +2946,10 @@ e1000_xmit_frame(struct sk_buff *skb, struct net_device *netdev) unsigned int f; unsigned int f; len -= skb->data_len; len -= skb->data_len; /* This goes back to the question of how to logically map a tx queue * to a flow. Right now, performance is impacted slightly negatively * if using multiple tx queues. If the stack breaks away from a * single qdisc implementation, we can look at this again. */ tx_ring = adapter->tx_ring; tx_ring = adapter->tx_ring; if (unlikely(skb->len <= 0)) { if (unlikely(skb->len <= 0)) { Loading Loading @@ -3012,8 +3045,7 @@ e1000_xmit_frame(struct sk_buff *skb, struct net_device *netdev) /* need: count + 2 desc gap to keep tail from touching /* need: count + 2 desc gap to keep tail from touching * head, otherwise try next time */ * head, otherwise try next time */ if (unlikely(E1000_DESC_UNUSED(tx_ring) < count + 2)) { if (unlikely(e1000_maybe_stop_tx(netdev, tx_ring, count + 2))) { netif_stop_queue(netdev); spin_unlock_irqrestore(&tx_ring->tx_lock, flags); spin_unlock_irqrestore(&tx_ring->tx_lock, flags); return NETDEV_TX_BUSY; return NETDEV_TX_BUSY; } } Loading Loading @@ -3060,8 +3092,7 @@ e1000_xmit_frame(struct sk_buff *skb, struct net_device *netdev) netdev->trans_start = jiffies; netdev->trans_start = jiffies; /* Make sure there is space in the ring for the next send. */ /* Make sure there is space in the ring for the next send. */ if (unlikely(E1000_DESC_UNUSED(tx_ring) < MAX_SKB_FRAGS + 2)) e1000_maybe_stop_tx(netdev, tx_ring, MAX_SKB_FRAGS + 2); netif_stop_queue(netdev); spin_unlock_irqrestore(&tx_ring->tx_lock, flags); spin_unlock_irqrestore(&tx_ring->tx_lock, flags); return NETDEV_TX_OK; return NETDEV_TX_OK; Loading Loading @@ -3556,13 +3587,14 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter, tx_ring->next_to_clean = i; tx_ring->next_to_clean = i; #define TX_WAKE_THRESHOLD 32 #define TX_WAKE_THRESHOLD 32 if (unlikely(cleaned && netif_queue_stopped(netdev) && if (unlikely(cleaned && netif_carrier_ok(netdev) && netif_carrier_ok(netdev))) { E1000_DESC_UNUSED(tx_ring) >= TX_WAKE_THRESHOLD)) { spin_lock(&tx_ring->tx_lock); /* Make sure that anybody stopping the queue after this if (netif_queue_stopped(netdev) && * sees the new next_to_clean. (E1000_DESC_UNUSED(tx_ring) >= TX_WAKE_THRESHOLD)) */ smp_mb(); if (netif_queue_stopped(netdev)) netif_wake_queue(netdev); netif_wake_queue(netdev); spin_unlock(&tx_ring->tx_lock); } } if (adapter->detect_tx_hung) { if (adapter->detect_tx_hung) { Loading