Re: [PATCH v2 1/4] tcp: fix too short FIN_WAIT2 time out

[Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>]

On Tue, 2013-03-19 at 09:32 -0400, David Miller wrote:
> From: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
> Date: Mon, 18 Mar 2013 21:39:04 +0900
> 
> >               if (tp->linger2 >= 0) {
> > -                     const int tmo = tcp_fin_time(sk) - TCP_TIMEWAIT_LEN;
> > -
> > -                     if (tmo > 0) {
> > -                             tcp_time_wait(sk, TCP_FIN_WAIT2, tmo);
> > -                             goto out;
> > -                     }
> > +                     tcp_time_wait(sk, TCP_FIN_WAIT2, TCP_TIMEWAIT_LEN);
> > +                     goto out;
> >               }
> 
> Well, now you're completely ignoring the user's linger setting.

If you mention TCP_LINGER2, I don't think I'm ignoring it.
It is taken into account in tcp_rcv_state_process() or tcp_close().
If I'm misunderstanding, I'd be glad if you could point out it.

> 
> I really can't take these patches seriously, and will not apply them,
> sorry.

I think, at least, too short timeout is harmful.
If tcp_fin_timeout is set to 61, it will expire in 2 seconds, which
cause peer to receive unexpected reset by sending fin.
Don't you think there is a problem?

Toshiaki Makita