Date: Thu, 2 Mar 95 14:26:51 MST From: terry@cs.weber.edu (Terry Lambert) To: tjohnson@cobber.cord.edu (Protius) Cc: questions@FreeBSD.org Subject: Re: gateway Message-ID: <9503022126.AA22003@cs.weber.edu> In-Reply-To: <9503022100.AA19438@cobber.cord.edu> from "Protius" at Mar 2, 95 03:00:28 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > We have a 386dx with a 387 running FreeBSD 2.1.0-Development (950210) > on a pair of ESDI drives, with a 3com 503 ethernet board. > > We're trying to get sliplogin running. So, we followed the FAQs, and > everything went great, until we put in the GATEWAY option and recompiled > the kernel. When it booted the new kernel, it tried to start the > networking stuff, and ifconfig, routed, and friends all said > "protocol not supported". Other demons, like named and httpd, just hung. > (by hitting ^C several times, I got a login prompt) > > One other interesting, but not related to this adventure, problem: > We tried to put a WD8013 card in this machine, and it would reliably > panic (page fault in kernel mode I think) when it tried to use the > ethernet card. This computer is an old AT&T 6386, we figured it was a > hardware incompatability. As an AT&T 6386 WGS, there are a lot of potential problems. Is this a 6386 WGS (25 MHz desktop), a 6386/E WGS (33MHz tower), or one of the personal systems that come by default with an IDE? The 25's used to come with WD7000-FASST SCSI interfaces, and the 33's used to come with a single 330M ESDI. My first working 386BSD 0.1 box was a 6386 25 with the WD1007 (jumpered for "perfect media") with the controller and ESDI drive from from one of the E's. There is a major big time problem with DMA on the 33's which kept every ethernet card we tried from working, except some old 3COM 3C503's. The AT&T patch for the AT&T NAU ethernet boards was to rewrite the memory contents, reading until it was correct, before triggering a send. This is basically the patch they gave to Davis County, which also has a bunch of 6386/E WGS's. The guy who wrote the original driver was Kurt Mahon (yes, the guy who used to live in the office exactly one floor below me at Novell, Sandy). He said that this was the only possible fix. Typically, I'd suggest 3COM cards to avoid the problem. Barring that, jumper an extra wait state into the thing and avoid the 0x3xx address range on your cards. Oh, and disabling the serial ports won't get them out of the way as far as address range goes. The things use the same Intel manufactured motherboards that made the Intel 302's (resold by Unisys on the Desktop III contract -- I worked on OpenNet for that contract) require the Ungerman-Bass boards to do networking at all. Sorry, I don't have any of the old notes that I took while beating the things into usability. There are currently two 6386/E's in service as FreeBSD boxes and one as a NetBSD box, and one 6386 25 as a 386BSD 0.1 + patchkit box, currently in service at Weber right now. All the 33's are running *ONLY* 3COM cards. Terry Lambert terry@cs.weber.edu --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9503022126.AA22003>