tcp: add one more quick ack after after ECN events
authorEric Dumazet <edumazet@google.com>
Wed, 27 Jun 2018 15:47:21 +0000 (08:47 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Fri, 3 Aug 2018 05:50:45 +0000 (07:50 +0200)
commit0a2f6725115dc17b97294403a5afca830c64e48a
tree74f4d2d5d6dcf5504379d8175423627761edea98
parent1f04d750f6688c0fe0cc13fd05d5744d1e7b50c3
tcp: add one more quick ack after after ECN events

[ Upstream commit 15ecbe94a45ef88491ca459b26efdd02f91edb6d ]

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>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
net/ipv4/tcp_input.c