*BSD News Article 19754


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!munnari.oz.au!network.ucsd.edu!ogicse!emory!europa.eng.gtefsd.com!uunet!email.tuwien.ac.at!ntkupc1.tuwien.ac.at!not-for-mail
From: martin@ntkupc1.tuwien.ac.at (Martin Birgmeier)
Newsgroups: comp.os.386bsd.questions
Subject: Re: [Q] com1: silo owerflow and slip
Message-ID: <251sdr$mkl@ntkupc1.tuwien.ac.at>
Date: 20 Aug 93 10:53:47 GMT
Article-I.D.: ntkupc1.251sdr$mkl
References: <1993Aug19.135931.165@ebc.ee> <250bki$r9m@pdq.coe.montana.edu> <hastyCC0rvp.63J@netcom.com>
Organization: Dept. of Comm. Engr., Tech. Univ. of Vienna, Austria
Lines: 39
NNTP-Posting-Host: ntkupc1.tuwien.ac.at

In article <hastyCC0rvp.63J@netcom.com>,
Amancio Hasty Jr <hasty@netcom.com> wrote:
>In article <250bki$r9m@pdq.coe.montana.edu> osyjm@cs.montana.edu
>(Jaye Mathisen) writes:
>>In article <1993Aug19.135931.165@ebc.ee>, Marek Tiits <marek@utlib.ee> wrote:
[...]
>>>	i've just successfully installed NetBSD 0.8 on ISA 486-33 with
>>>8 MB RAM and 350MB Conner IDE disk. BUT yhe problem is, that my 9600bps
>>>leased line slip connection is very unstable. I'm geting messages
>>>
>>>	com1: silo overflow
[...]
>>It is odd that you're having problems at 9600 baud, I would'nt have guessed
>>you'd have trouble on a 486 with 9600 baud.
[...]
>Also, does the com port or modem have 16550 hardware buffer?
>If not, you will get com overflows...
[...]
Actually, I think the real cause is a bad serial board (or some related
hardware problem)... I used to have a board with 2 * 16450 in my machine;
usually, for a time after power-on, everything would work fine, but soon
later I would get overflows all the time, this getting so bad that the
X server (and especially the window manager) got so confused that it (the
latter) locked up. So it might also be a thermal problem. One more thing -
overflows also seemed to increase during (SCSI) disk activity. I changed the
board against a similar one, with exactly the same problems. Finally
I changed it against a very old board (1 * 8250, 1 * 16450), and I have
never had any problems since then! Of course, there might still be a
problem with the software, maybe it's configuring the 16450 incorrectly
whereas the 8250 gets configured ok, but I don't think this is the reason
since the single 16450 on my current board behaves well (the mouse is
connected to the 8250).

Hope this helps,

- Martin
-- 
Martin Birgmeier
Technical University of Vienna                      mbirgmei@email.tuwien.ac.at