net/mlx4_en: Fix a race when closing TX queue

There is a possible race where the TX completion handler can clean the
entire TX queue between the decision that the queue is full and actually
closing it. To avoid this situation, check again if the queue is really
full, if not, reopen the transmit and continue with sending the packet.

CC: Eric Dumazet <edumazet@google.com>
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.com>
Signed-off-by: Eugenia Emantayev <eugenia@mellanox.com>
Signed-off-by: Amir Vadai <amirv@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/drivers/net/ethernet/mellanox/mlx4/en_tx.c b/drivers/net/ethernet/mellanox/mlx4/en_tx.c
index 16af338..3c5ffd2 100644 (file)
--- a/drivers/net/ethernet/mellanox/mlx4/en_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_tx.c
@@ -588,7 +588,21 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
                netif_tx_stop_queue(ring->tx_queue);
                priv->port_stats.queue_stopped++;
 
-               return NETDEV_TX_BUSY;
+               /* If queue was emptied after the if, and before the
+                * stop_queue - need to wake the queue, or else it will remain
+                * stopped forever.
+                * Need a memory barrier to make sure ring->cons was not
+                * updated before queue was stopped.
+                */
+               wmb();
+
+               if (unlikely(((int)(ring->prod - ring->cons)) <=
+                            ring->size - HEADROOM - MAX_DESC_TXBBS)) {
+                       netif_tx_wake_queue(ring->tx_queue);
+                       priv->port_stats.wake_queue++;
+               } else {
+                       return NETDEV_TX_BUSY;
+               }
        }
 
        /* Track current inflight packets for performance analysis */