lkml.org 
[lkml]   [1999]   [Mar]   [15]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: No connect timeout?
Date
Hello!

> > Now still to find the pkt_too_big to self and copied < acked problems
> > Jason reported. Alexey, do you have a theory for the pkt_too_big to
> > self?
>
> I am still getting massive numbers of that message. The tcp_data message I
> have not seen anymore which is odd, it was very frequent two days ago.
> With the extra print outs that Andi asked for:
>
> sending pkt_too_big to self totlen=1500,pmtu=576
> sending pkt_too_big to self totlen=796,pmtu=296
> sending pkt_too_big to self totlen=1500,pmtu=296

If connection does not die after this, it is harmless.

It is natural race between dst->pmtu and sendmsg (which is amplified,
when socket starts to send big frames when cwnd>1 and with space in sndbuf).
I remember, we decided to NETDEBUG() this. Certainly, we could insert recalc.
of mss everywhere in sender path, but it does not remove the race in any case,
but will introcude new bits in critical path.

What's about copied<acked, it is more serious, I used to believe
that they are observed only in 2.0. No ideas here.

Alexey

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:50    [W:0.129 / U:0.160 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site