From owner-freebsd-hackers Wed Jun 12 08:26:52 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id IAA12017 for hackers-outgoing; Wed, 12 Jun 1996 08:26:52 -0700 (PDT) Received: from Glock.COM (root@glock.com [198.82.228.165]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id IAA12011 for ; Wed, 12 Jun 1996 08:26:47 -0700 (PDT) Received: (from mmead@localhost) by Glock.COM (8.7.3/8.7.3) id LAA01393; Wed, 12 Jun 1996 11:21:40 -0400 (EDT) From: "matthew c. mead" Message-Id: <199606121521.LAA01393@Glock.COM> Subject: Re: ST43400N To: se@zpr.uni-koeln.de (Stefan Esser) Date: Wed, 12 Jun 1996 11:21:40 -0400 (EDT) Cc: hackers@freebsd.org In-Reply-To: <199606112010.AA13884@Sisyphos> from "Stefan Esser" at Jun 11, 96 10:10:41 pm X-Mailer: ELM [version 2.4 PL24 ME8a] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Stefan Esser writes: Well, I know this is a second reply, but while I was asleep last night the thing started doing those ccb timeouts again. The system still worked, but it took me 5 minutes to log in and reboot it the kernel was so busy with the disk errors. I'm gonna try some of the sugeestions below: > } > If you are running a -current kernel, then please try > } > one built with "options FAILSAFE". This will disable > } > use of tagged command queues, which are the most likely > } > cause of the problem you report. > } > If the drive works with tags disabled, then I'll send > } > further directions to find out whether it is possible > } > to work around that problem. (You may comment out the > } > call to scsi_start_unit() in /sys/scsi/sd.c for a test.) > } Do the above two paragraphs apply to a 2.1.0-RELEASE > } kernel as well? This drive is on a 2.1.0-R system. > Look at /sys/pci/ncr.c (line 6242 in -stable): > ncr_setmaxtags (tp, SCSI_NCR_MAX_TAGS); > Change this into: > ncr_setmaxtags (tp, 0); > This is part of what FAILSAFE does in current kernels ... Ok, I did this. > The paragraph about scsi_start_unit() applies to 2.1.0R. If things don't work this way I'll remove that next. Thanks for the suggestions. -matt -- Matthew C. Mead mmead@goof.com http://www.goof.com/~mmead/