On Sun, 2017-05-21 at 13:51 -0400, David Miller wrote: > From: Eric Dumazet <eric.duma...@gmail.com> > Date: Sun, 21 May 2017 10:39:00 -0700 > > > From: Eric Dumazet <eduma...@google.com> > > > > TCP_USER_TIMEOUT is still converted to jiffies value in > > icsk_user_timeout > > > > So we need to make a conversion for the cases HZ != 1000 > > > > Fixes: 9a568de4818d ("tcp: switch TCP TS option (RFC 7323) to 1ms clock") > > Signed-off-by: Eric Dumazet <eduma...@google.com> > > Applied, thank Eric. > > I kinda expected a few pieces of fallout from the 1ms changes :)
Absolutely ;) One last piece is in TCP_SYNCNT support. I saw that retransmits_timed_out() could have a rounding error : tcp_time_stamp(tcp_sk(sk)) - start_ts) ends up to 999 ms, while the timeout is/was 1000 ms (And timer _was_ progammed with 1000 jiffies for HZ=1000 kernel) So if user setup TCP_SYNCNT = 1 socket option, we sometime sends one extra SYN packet. I will send a fix later.