[PATCH net] tcp: add one more quick ack after after ECN events

[PATCH net] tcp: add one more quick ack after after ECN events

[Eric Dumazet]

Larry Brakmo proposal ( https://patchwork.ozlabs.org/patch/935233/
tcp: force cwnd at least 2 in tcp_cwnd_reduction) made us rethink
about our recent patch removing ~16 quick acks after ECN events.

tcp_enter_quickack_mode(sk, 1) makes sure one immediate ack is sent,
but in the case the sender cwnd was lowered to 1, we do not want
to have a delayed ack for the next packet we will receive.

Fixes: 522040ea5fdd ("tcp: do not aggressively quick ack after ECN events")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Neal Cardwell <ncardwell@google.com>
Cc: Lawrence Brakmo <brakmo@fb.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 355d3dffd021ccad0f30891994289d916f7d276c..045d930d01a92c2b0b7535062b32c27f6c150460 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -265,7 +265,7 @@ static void __tcp_ecn_check_ce(struct sock *sk, const struct sk_buff *skb)
 		 * it is probably a retransmit.
 		 */
 		if (tp->ecn_flags & TCP_ECN_SEEN)
-			tcp_enter_quickack_mode(sk, 1);
+			tcp_enter_quickack_mode(sk, 2);
 		break;
 	case INET_ECN_CE:
 		if (tcp_ca_needs_ecn(sk))
@@ -273,7 +273,7 @@ static void __tcp_ecn_check_ce(struct sock *sk, const struct sk_buff *skb)
 
 		if (!(tp->ecn_flags & TCP_ECN_DEMAND_CWR)) {
 			/* Better not delay acks, sender can have a very low cwnd */
-			tcp_enter_quickack_mode(sk, 1);
+			tcp_enter_quickack_mode(sk, 2);
 			tp->ecn_flags |= TCP_ECN_DEMAND_CWR;
 		}
 		tp->ecn_flags |= TCP_ECN_SEEN;
-- 
2.18.0.rc2.346.g013aa6912e-goog

Re: [PATCH net] tcp: add one more quick ack after after ECN events

[Neal Cardwell]

On Wed, Jun 27, 2018 at 11:47 AM Eric Dumazet <edumazet@google.com> wrote:
>
> Larry Brakmo proposal ( https://patchwork.ozlabs.org/patch/935233/
> tcp: force cwnd at least 2 in tcp_cwnd_reduction) made us rethink
> about our recent patch removing ~16 quick acks after ECN events.
>
> tcp_enter_quickack_mode(sk, 1) makes sure one immediate ack is sent,
> but in the case the sender cwnd was lowered to 1, we do not want
> to have a delayed ack for the next packet we will receive.
>
> Fixes: 522040ea5fdd ("tcp: do not aggressively quick ack after ECN events")
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Reported-by: Neal Cardwell <ncardwell@google.com>
> Cc: Lawrence Brakmo <brakmo@fb.com>
> ---
>  net/ipv4/tcp_input.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)

Acked-by: Neal Cardwell <ncardwell@google.com>

Thanks, Eric!

neal

Re: [PATCH net] tcp: add one more quick ack after after ECN events

[David Miller]

From: Eric Dumazet <edumazet@google.com>
Date: Wed, 27 Jun 2018 08:47:21 -0700

> Larry Brakmo proposal ( https://patchwork.ozlabs.org/patch/935233/
> tcp: force cwnd at least 2 in tcp_cwnd_reduction) made us rethink
> about our recent patch removing ~16 quick acks after ECN events.
> 
> tcp_enter_quickack_mode(sk, 1) makes sure one immediate ack is sent,
> but in the case the sender cwnd was lowered to 1, we do not want
> to have a delayed ack for the next packet we will receive.
> 
> Fixes: 522040ea5fdd ("tcp: do not aggressively quick ack after ECN events")
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Reported-by: Neal Cardwell <ncardwell@google.com>
> Cc: Lawrence Brakmo <brakmo@fb.com>

Applied.