[PATCH net] tcp: avoid setting cwnd to invalid ssthresh after cwnd reduction states

If the sender switches the congestion control during ECN-triggered
cwnd-reduction state (CA_CWR), upon exiting recovery cwnd is set to
the ssthresh value calculated by the previous congestion control. If
the previous congestion control is BBR that always keep ssthresh
to TCP_INIFINITE_SSTHRESH, cwnd ends up being infinite. The safe
step is to avoid assigning invalid ssthresh value when recovery ends.

Signed-off-by: Yuchung Cheng <redacted>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
---
 net/ipv4/tcp_input.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 2920e0cb09f8..dad026fcfd09 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c

@@ -2520,8 +2520,8 @@ static inline void tcp_end_cwnd_reduction(struct sock *sk)
 		return;
 
 	/* Reset cwnd to ssthresh in CWR or Recovery (unless it's undone) */
-	if (inet_csk(sk)->icsk_ca_state == TCP_CA_CWR ||
-	    (tp->undo_marker && tp->snd_ssthresh < TCP_INFINITE_SSTHRESH)) {
+	if (tp->snd_ssthresh < TCP_INFINITE_SSTHRESH &&
+	    (inet_csk(sk)->icsk_ca_state == TCP_CA_CWR || tp->undo_marker)) {
 		tp->snd_cwnd = tp->snd_ssthresh;
 		tp->snd_cwnd_stamp = tcp_jiffies32;
 	}

-- 
2.14.0.rc1.383.gd1ce394fe2-goog

On Tue, 2017-08-01 at 13:22 -0700, Yuchung Cheng wrote:

If the sender switches the congestion control during ECN-triggered
cwnd-reduction state (CA_CWR), upon exiting recovery cwnd is set to
the ssthresh value calculated by the previous congestion control. If
the previous congestion control is BBR that always keep ssthresh
to TCP_INIFINITE_SSTHRESH, cwnd ends up being infinite. The safe
step is to avoid assigning invalid ssthresh value when recovery ends.

Signed-off-by: Yuchung Cheng <redacted>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
---
 net/ipv4/tcp_input.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Acked-by: Eric Dumazet <redacted>

From: Yuchung Cheng <redacted>
Date: Tue,  1 Aug 2017 13:22:32 -0700

If the sender switches the congestion control during ECN-triggered
cwnd-reduction state (CA_CWR), upon exiting recovery cwnd is set to
the ssthresh value calculated by the previous congestion control. If
the previous congestion control is BBR that always keep ssthresh
to TCP_INIFINITE_SSTHRESH, cwnd ends up being infinite. The safe
step is to avoid assigning invalid ssthresh value when recovery ends.

Signed-off-by: Yuchung Cheng <redacted>
Signed-off-by: Neal Cardwell <ncardwell@google.com>

Applied, thanks.

Is this a -stable candidate?

On Wed, Aug 2, 2017 at 10:51 AM, David Miller [off-list ref] wrote:

From: Yuchung Cheng <redacted>
Date: Tue,  1 Aug 2017 13:22:32 -0700

quoted

If the sender switches the congestion control during ECN-triggered
cwnd-reduction state (CA_CWR), upon exiting recovery cwnd is set to
the ssthresh value calculated by the previous congestion control. If
the previous congestion control is BBR that always keep ssthresh
to TCP_INIFINITE_SSTHRESH, cwnd ends up being infinite. The safe
step is to avoid assigning invalid ssthresh value when recovery ends.

Signed-off-by: Yuchung Cheng <redacted>
Signed-off-by: Neal Cardwell <ncardwell@google.com>

Applied, thanks.

Is this a -stable candidate?

Yes it is. Thanks!

From: Yuchung Cheng <redacted>
Date: Wed, 2 Aug 2017 10:58:54 -0700

On Wed, Aug 2, 2017 at 10:51 AM, David Miller [off-list ref] wrote:

quoted

From: Yuchung Cheng <redacted>
Date: Tue,  1 Aug 2017 13:22:32 -0700

quoted

If the sender switches the congestion control during ECN-triggered
cwnd-reduction state (CA_CWR), upon exiting recovery cwnd is set to
the ssthresh value calculated by the previous congestion control. If
the previous congestion control is BBR that always keep ssthresh
to TCP_INIFINITE_SSTHRESH, cwnd ends up being infinite. The safe
step is to avoid assigning invalid ssthresh value when recovery ends.

Signed-off-by: Yuchung Cheng <redacted>
Signed-off-by: Neal Cardwell <ncardwell@google.com>

Applied, thanks.

Is this a -stable candidate?

Yes it is. Thanks!

Great, queued up.