Date: Wed, 8 Oct 1997 21:50:18 +0200 From: j@uriah.heep.sax.de (J Wunsch) To: hackers@FreeBSD.ORG Cc: mrcpu@cdsnet.net (Jaye Mathisen) Subject: Re: A TCP problem, plus solution! (3.0-current) Message-ID: <19971008215018.SO62644@uriah.heep.sax.de> In-Reply-To: <Pine.NEB.3.95.971008104120.25558C-100000@mail.cdsnet.net>; from Jaye Mathisen on Oct 8, 1997 10:44:28 -0700 References: <Pine.NEB.3.95.971008104120.25558C-100000@mail.cdsnet.net>
next in thread | previous in thread | raw e-mail | index | archive | help
As Jaye Mathisen wrote: > I traced it to a sysctl I had in /etc/rc.local that bumped the tcp > recvspace to 65536. > > Resetting it to the default clears the problem up. You are not running -current. ;-) RCS file: /home/cvs/src/sys/netinet/tcp_output.c,v ... revision 1.27 date: 1997/10/07 21:10:06; author: fenner; state: Exp; lines: +3 -3 Don't allow the window to be increased beyond what is possible to represent in the TCP header. The old code did effectively: win = min(win, MAX_ALLOWED); win = max(win, what_i_think_i_advertised_last_time); so if what_i_think_i_advertised_last_time is bigger than can be represented in the header (e.g. large buffers and no window scaling) then we stuff a too-big number into a short. This fix reverses the order of the comparisons. PR: kern/4712 -- cheers, J"org joerg_wunsch@uriah.heep.sax.de -- http://www.sax.de/~joerg/ -- NIC: JW11-RIPE Never trust an operating system you don't have sources for. ;-)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19971008215018.SO62644>