*BSD News Article 38753


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!newshost.marcam.com!zip.eecs.umich.edu!panix!ddsw1!not-for-mail
From: chilton@MCS.COM (Christopher Hilton)
Newsgroups: comp.os.386bsd.questions
Subject: Re: FreeBSD and Adaptec 1542CF SCSI
Date: 2 Dec 1994 09:59:32 -0600
Organization: /usr/lib/news/organi[sz]ation
Lines: 59
Message-ID: <3bng94$iaf@Mars.mcs.com>
References: <3b2u2e$27u@scoob.xap.com> <Roy-2711942134250001@adept.cts.com> <9412012201.35@rmkhome.com>
NNTP-Posting-Host: mars.mcs.com

In article <9412012201.35@rmkhome.com>, Rick Kelly <rmk@rmkhome.com> wrote:
>Roy Lovejoy (Roy@AdeptSolutions.com) wrote:
>: In article <3b2u2e$27u@scoob.xap.com>, nb@xap.com (Nathan Bradshaw) wrote:
>: > My roommate installed 2.0 RELEASE yesterday on his new Pentium 90 VLB/PCI
>: > 16Mb with an ISA Adaptec 1542cf and a Conner 1.2Gb disk.  Overnight it
>: > bombed out with an error much like:

[snip]

> 
>I have 2 systems here.  One is a NetBSD system and the other is Coherent.
>
>The Coherent box is a 486DX33 with 16 megs of memory and an Adaptec 1542cf
>with about 9 feet of SCSI cable.  It is a news, mail, and development box
>that grinds away 24 hours a day, 7 days a week, with nary a timeout from
>the Adaptec.  I have the Coherent driver sources, and the SCSI driver doesn't
>seem to do any special dance with a 1542cf.

I'm glad to see that someone is using my stuff so successfully.

>
>Maybe it's a termination problem or some SCSI peripheral is wedging the
>SCSI bus.


I repeat: The early 1542C and CF's had really cheesy SCSI linde
drivers (hardware) and as a result are really affected by cable length
and type. I don't have the SCSI spec in from of me anymoe (not my job
anymore) But I seem to remember the spec saying something like 3
meters is the maximum cable length. Stick with that. If you're stuck
in the English system of measurement 3 meters == 9' 10.11" or closer to
10' then 9'. Rick, your cable is probably within spec.

>
>I have used the 154xA and 154xB controllers with 15 foot, flat, unshielded
>SCSI cables without the problems above.
>

The 1542A and B controllers had much better (and much more expensive)
discrete line drivers and weren't genereally subject to these
problems. Talk to old-hands in comp.periphs.scsi for this one. As a
matter of fact I think that it's still in the FAQ.

Rick's suggestions that you check termination etc. are valid BTW. My
experience has been that DOS is much more tolerant of these types of
problems than is Unix which stresses the system and SCSI bus much
more. Also don't forget that your system tries to fsck all the disks
at night which is going to put a load on the system that might trigger
your problem.

	-Chris Hilton
-- 
----------------------------------------------------------------------
Christopher Sean Hilton	| "Thus it is said if you know them and know 
<chilton@mcs.com>       | yourself, your victory will not be imperiled.
For PGP key finger:     | If you know Heaven and you know Earth, your 
     chilton@mcs.com    | victory will be complete." - Sun Tsu
----------------------------------------------------------------------