Subject: Is this a bug in tcp_input.c ? (fwd)
To: None <tech-kern@NetBSD.ORG>
From: Darren Reed <firstname.lastname@example.org>
Date: 05/17/1995 18:26:28
>From comp.protocols.tcp-ip, I found the following:
In article <email@example.com>, firstname.lastname@example.org (Rick Jones) writes:
|> James Carlson (carlson@Xylogics.COM) wrote:
|> : In article <3oms03$j1v@cnn.Princeton.EDU>, email@example.com (David A. Wagner) writes:
|> : [...]
|> : |> iss = tp->rcv_nxt + TCP_ISSINCR;
|> : |>
|> : |> Shouldn't that be a tp->snd_nxt instead of a tp->rcv_nxt ?
|> : [...]
|> : Yep, sure looks like it to me ...
|> I thought that what we are checking is the SYN from the remote to make
|> sure that it is beyond the window, beyond the window would be beyond
|> what we ahve been *receiving* from the remote, so wouldn't it be
Yeah, the part of the code above where I extracted (the 'if' statement
itself) is fine. But iss shouldn't be set to the rcv_nxt plus a small
increment -- iss is the *sender's* initial sequence number, not the
expected (received) sequence number.
James Carlson <firstname.lastname@example.org> Tel: +1 617 272 8140
Annex Software Support / Xylogics, Inc. +1 800 225 3317
53 Third Avenue / Burlington MA 01803-4491 Fax: +1 617 272 2618