From: Eric Dumazet Date: Thu, 29 Aug 2013 01:10:43 +0000 (-0700) Subject: net: revert 8728c544a9c ("net: dev_pick_tx() fix") X-Git-Url: http://git.mmlx.us/?a=commitdiff_plain;h=56a12acebcbd08342f7287a5870fe7ec2c0de91a;p=linux-edison.git net: revert 8728c544a9c ("net: dev_pick_tx() fix") [ Upstream commit 702821f4ea6f68db18aa1de7d8ed62c6ba586a64 ] commit 8728c544a9cbdc ("net: dev_pick_tx() fix") and commit b6fe83e9525a ("bonding: refine IFF_XMIT_DST_RELEASE capability") are quite incompatible : Queue selection is disabled because skb dst was dropped before entering bonding device. This causes major performance regression, mainly because TCP packets for a given flow can be sent to multiple queues. This is particularly visible when using the new FQ packet scheduler with MQ + FQ setup on the slaves. We can safely revert the first commit now that 416186fbf8c5b ("net: Split core bits of netdev_pick_tx into __netdev_pick_tx") properly caps the queue_index. Reported-by: Xi Wang Diagnosed-by: Xi Wang Signed-off-by: Eric Dumazet Cc: Tom Herbert Cc: Alexander Duyck Cc: Denys Fedorysychenko Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman --- diff --git a/net/core/flow_dissector.c b/net/core/flow_dissector.c index 00ee068efc1..c99cc371bbd 100644 --- a/net/core/flow_dissector.c +++ b/net/core/flow_dissector.c @@ -345,14 +345,9 @@ u16 __netdev_pick_tx(struct net_device *dev, struct sk_buff *skb) if (new_index < 0) new_index = skb_tx_hash(dev, skb); - if (queue_index != new_index && sk) { - struct dst_entry *dst = - rcu_dereference_check(sk->sk_dst_cache, 1); - - if (dst && skb_dst(skb) == dst) - sk_tx_queue_set(sk, queue_index); - - } + if (queue_index != new_index && sk && + rcu_access_pointer(sk->sk_dst_cache)) + sk_tx_queue_set(sk, queue_index); queue_index = new_index; }