tcp: switch TCP TS option (RFC 7323) to 1ms clock
authorEric Dumazet <edumazet@google.com>
Tue, 16 May 2017 21:00:14 +0000 (14:00 -0700)
committerDavid S. Miller <davem@davemloft.net>
Wed, 17 May 2017 20:06:01 +0000 (16:06 -0400)
commit9a568de4818dea9a05af141046bd3e589245ab83
tree6f1502edf55ecb7205660d62bd683ebcf912cfea
parentac9517fcf310327fa3e3b0d8366e4b11236b1b4b
tcp: switch TCP TS option (RFC 7323) to 1ms clock

TCP Timestamps option is defined in RFC 7323

Traditionally on linux, it has been tied to the internal
'jiffies' variable, because it had been a cheap and good enough
generator.

For TCP flows on the Internet, 1 ms resolution would be much better
than 4ms or 10ms (HZ=250 or HZ=100 respectively)

For TCP flows in the DC, Google has used usec resolution for more
than two years with great success [1]

Receive size autotuning (DRS) is indeed more precise and converges
faster to optimal window size.

This patch converts tp->tcp_mstamp to a plain u64 value storing
a 1 usec TCP clock.

This choice will allow us to upstream the 1 usec TS option as
discussed in IETF 97.

[1] https://www.ietf.org/proceedings/97/slides/slides-97-tcpm-tcp-options-for-low-latency-00.pdf

Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 files changed:
include/linux/skbuff.h
include/linux/tcp.h
include/net/tcp.h
net/ipv4/syncookies.c
net/ipv4/tcp.c
net/ipv4/tcp_bbr.c
net/ipv4/tcp_input.c
net/ipv4/tcp_ipv4.c
net/ipv4/tcp_lp.c
net/ipv4/tcp_minisocks.c
net/ipv4/tcp_output.c
net/ipv4/tcp_rate.c
net/ipv4/tcp_recovery.c
net/ipv4/tcp_timer.c
net/ipv6/syncookies.c
net/ipv6/tcp_ipv6.c
net/netfilter/nf_synproxy_core.c