From: Yuchung Cheng <ycheng@google.com>
Date: Sun, 24 Mar 2013 10:42:25 +0000 (+0000)
Subject: tcp: undo spurious timeout after SACK reneging
X-Git-Tag: v3.2.43~20
X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=4289cdeb48d9d5d2674a85e63b315d4771bf4d27;p=thirdparty%2Fkernel%2Fstable.git

tcp: undo spurious timeout after SACK reneging

[ Upstream commit 7ebe183c6d444ef5587d803b64a1f4734b18c564 ]

On SACK reneging the sender immediately retransmits and forces a
timeout but disables Eifel (undo). If the (buggy) receiver does not
drop any packet this can trigger a false slow-start retransmit storm
driven by the ACKs of the original packets. This can be detected with
undo and TCP timestamps.

Signed-off-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
---

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 1b1f7af17d6a6..3124e17675f95 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -2265,11 +2265,8 @@ void tcp_enter_loss(struct sock *sk, int how)
 	if (tcp_is_reno(tp))
 		tcp_reset_reno_sack(tp);
 
-	if (!how) {
-		/* Push undo marker, if it was plain RTO and nothing
-		 * was retransmitted. */
-		tp->undo_marker = tp->snd_una;
-	} else {
+	tp->undo_marker = tp->snd_una;
+	if (how) {
 		tp->sacked_out = 0;
 		tp->fackets_out = 0;
 	}