From owner-freebsd-current Thu Apr 4 11:21:54 1996 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id LAA21834 for current-outgoing; Thu, 4 Apr 1996 11:21:54 -0800 (PST) Received: from brasil.moneng.mei.com (brasil.moneng.mei.com [151.186.109.160]) by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id LAA21829 for ; Thu, 4 Apr 1996 11:21:50 -0800 (PST) Received: (from jgreco@localhost) by brasil.moneng.mei.com (8.7.Beta.1/8.7.Beta.1) id NAA01863; Thu, 4 Apr 1996 13:19:26 -0600 From: Joe Greco Message-Id: <199604041919.NAA01863@brasil.moneng.mei.com> Subject: Re: tty-level buffer overflows - what to do? To: nate@sri.MT.net (Nate Williams) Date: Thu, 4 Apr 1996 13:19:26 -0600 (CST) Cc: jgreco@brasil.moneng.mei.com, root@deadline.snafu.de, current@FreeBSD.org In-Reply-To: <199604041912.MAA17250@rocky.sri.MT.net> from "Nate Williams" at Apr 4, 96 12:12:03 pm X-Mailer: ELM [version 2.4 PL24] Content-Type: text Sender: owner-current@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk > > My gut instinct is that you would find a correlation between your IDE disk > > going and these error messages. A 486/40 should be adequate, even for > > several lines at 115200. The fact that you are tight on memory would tend > > to cause you to hit the disk correspondingly more often, which would cause > > some burps in serial I/O... the fact that you're running rlogin also would > > tend to cause you to swap more, if you have a few active sessions. One of > > the nice things about kernel-mode SLIP is that no paging is involved.... > > That's possible, but let me throw in another data point. > > 486/66 - 16MB memory > 3 16550 UART serial lines - 115K > 540MB IDE drive > > With 2 lines going full blast (sup updates!) I see *NO* overflows on my > box. Again, my box is running with about 6MB of free memory all the > time, so I rarely hit the disk, but I've done compiles on the machine to > upgrade software with no noticeable degradation of serial speed. > However, I do no paging at all, so my disk is mostly idle even during > compiles. Run "du /" and watch your serial comms get choppy :-/ If you are running kernel mode SLIP or PPP, remember that you have several additional advantages over the described configuration: 1) the code to deal with the connection can't get swapped out. 2) no context switch overhead.. 3) you don't have to go through all the tty processing layers. with only 4MB of memory, the described configuration is not likely to have "free memory". ... Joe ------------------------------------------------------------------------------- Joe Greco - Systems Administrator jgreco@ns.sol.net Solaria Public Access UNIX - Milwaukee, WI 414/546-7968