tcp: switch snt_synack back to measuring transmit time of first SYNACK
authorNeal Cardwell <ncardwell@google.com>
Mon, 30 Jun 2014 19:09:49 +0000 (15:09 -0400)
committerDavid S. Miller <davem@davemloft.net>
Tue, 8 Jul 2014 02:26:37 +0000 (19:26 -0700)
commit86c6a2c75ab97fe31844985169e26aea335432f9
treee3d15ca2ac9a89fbd5399ac6a1886822176fc392
parent0b88e7042a221a0318d726017b0f97aa42066826
tcp: switch snt_synack back to measuring transmit time of first SYNACK

Always store in snt_synack the time at which the server received the
first client SYN and attempted to send the first SYNACK.

Recent commit aa27fc501 ("tcp: tcp_v[46]_conn_request: fix snt_synack
initialization") resolved an inconsistency between IPv4 and IPv6 in
the initialization of snt_synack. This commit brings back the idea
from 843f4a55e (tcp: use tcp_v4_send_synack on first SYN-ACK), which
was going for the original behavior of snt_synack from the commit
where it was added in 9ad7c049f0f79 ("tcp: RFC2988bis + taking RTT
sample from 3WHS for the passive open side") in v3.1.

In addition to being simpler (and probably a tiny bit faster),
unconditionally storing the time of the first SYNACK attempt has been
useful because it allows calculating a performance metric quantifying
how long it took to establish a passive TCP connection.

Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Cc: Octavian Purdila <octavian.purdila@intel.com>
Cc: Jerry Chu <hkchu@google.com>
Acked-by: Octavian Purdila <octavian.purdila@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
include/net/tcp.h
net/ipv4/tcp_ipv4.c
net/ipv6/tcp_ipv6.c