netem: add limitation to reordered packets
authorEric Dumazet <edumazet@google.com>
Tue, 3 Jul 2012 20:55:21 +0000 (20:55 +0000)
committerDavid S. Miller <davem@davemloft.net>
Mon, 9 Jul 2012 07:01:49 +0000 (00:01 -0700)
commit960fb66e520a405dde39ff883f17ff2669c13d85
treec59b298e45158401419fc42088dad8edf754e535
parentb94e52f62683dc0b00c6d1b58b80929a078c0fd5
netem: add limitation to reordered packets

Fix two netem bugs :

1) When a frame was dropped by tfifo_enqueue(), drop counter
   was incremented twice.

2) When reordering is triggered, we enqueue a packet without
   checking queue limit. This can OOM pretty fast when this
   is repeated enough, since skbs are orphaned, no socket limit
   can help in this situation.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Mark Gordon <msg@google.com>
Cc: Andreas Terzis <aterzis@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Hagen Paul Pfeifer <hagen@jauu.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/sched/sch_netem.c