[ 3.5.y.z extended stable ] Patch "tcp: frto should not set snd_cwnd to 0" has been added to staging queue

[Herton Ronaldo Krzesinski]

This is a note to let you know that I have just added a patch titled

    tcp: frto should not set snd_cwnd to 0

to the linux-3.5.y-queue branch of the 3.5.y.z extended stable tree 
which can be found at:

 http://kernel.ubuntu.com/git?p=ubuntu/linux.git;a=shortlog;h=refs/heads/linux-3.5.y-queue

If you, or anyone else, feels it should not be added to this tree, please 
reply to this email.

For more information about the 3.5.y.z tree, see
https://wiki.ubuntu.com/Kernel/Dev/ExtendedStable

Thanks.
-Herton

------

>From a9a83407f4f4ad682032d2f542d2d69775c146fa Mon Sep 17 00:00:00 2001
From: Eric Dumazet <edumazet at google.com>
Date: Sun, 3 Feb 2013 09:13:05 +0000
Subject: [PATCH] tcp: frto should not set snd_cwnd to 0
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

commit 2e5f421211ff76c17130b4597bc06df4eeead24f upstream.

Commit 9dc274151a548 (tcp: fix ABC in tcp_slow_start())
uncovered a bug in FRTO code :
tcp_process_frto() is setting snd_cwnd to 0 if the number
of in flight packets is 0.

As Neal pointed out, if no packet is in flight we lost our
chance to disambiguate whether a loss timeout was spurious.

We should assume it was a proper loss.

Reported-by: Pasi Kärkkäinen <pasik at iki.fi>
Signed-off-by: Neal Cardwell <ncardwell at google.com>
Signed-off-by: Eric Dumazet <edumazet at google.com>
Cc: Ilpo Järvinen <ilpo.jarvinen at helsinki.fi>
Cc: Yuchung Cheng <ycheng at google.com>
Signed-off-by: David S. Miller <davem at davemloft.net>
Signed-off-by: Herton Ronaldo Krzesinski <herton.krzesinski at canonical.com>
---
 net/ipv4/tcp_input.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 19c430c..6affa92 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -3695,7 +3695,8 @@ static bool tcp_process_frto(struct sock *sk, int flag)
 	    ((tp->frto_counter >= 2) && (flag & FLAG_RETRANS_DATA_ACKED)))
 		tp->undo_marker = 0;

-	if (!before(tp->snd_una, tp->frto_highmark)) {
+	if (!before(tp->snd_una, tp->frto_highmark) ||
+	    !tcp_packets_in_flight(tp)) {
 		tcp_enter_frto_loss(sk, (tp->frto_counter == 1 ? 2 : 3), flag);
 		return true;
 	}
--
1.7.9.5