sch_sfq: avoid giving spurious NET_XMIT_CN signals
authorEric Dumazet <eric.dumazet@gmail.com>
Mon, 23 May 2011 11:02:42 +0000 (11:02 +0000)
committerGreg Kroah-Hartman <gregkh@suse.de>
Fri, 3 Jun 2011 00:31:57 +0000 (09:31 +0900)
commit9df49f2bfe862573911a080c75a6d81113c5c81d
treefcb24fb5990cbf43f37db714728b1a4a99dad5a8
parent37d37d7502216a7d0e0fb33018afd8e714963b20
sch_sfq: avoid giving spurious NET_XMIT_CN signals

[ Upstream commit 8efa885406359af300d46910642b50ca82c0fe47 ]

While chasing a possible net_sched bug, I found that IP fragments have
litle chance to pass a congestioned SFQ qdisc :

- Say SFQ qdisc is full because one flow is non responsive.
- ip_fragment() wants to send two fragments belonging to an idle flow.
- sfq_enqueue() queues first packet, but see queue limit reached :
- sfq_enqueue() drops one packet from 'big consumer', and returns
NET_XMIT_CN.
- ip_fragment() cancel remaining fragments.

This patch restores fairness, making sure we return NET_XMIT_CN only if
we dropped a packet from the same flow.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Patrick McHardy <kaber@trash.net>
CC: Jarek Poplawski <jarkao2@gmail.com>
CC: Jamal Hadi Salim <hadi@cyberus.ca>
CC: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
net/sched/sch_sfq.c