From owner-freebsd-current Sun Mar 5 03:31:45 1995 Return-Path: current-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id DAA03362 for current-outgoing; Sun, 5 Mar 1995 03:31:45 -0800 Received: from hda.com (hda.com [199.232.40.182]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id DAA03356 for ; Sun, 5 Mar 1995 03:31:42 -0800 Received: (dufault@localhost) by hda.com (8.6.9/8.3) id GAA01718; Sun, 5 Mar 1995 06:29:03 -0500 From: Peter Dufault Message-Id: <199503051129.GAA01718@hda.com> Subject: Re: Wiring To: hsu@cs.hut.fi Date: Sun, 5 Mar 1995 06:29:02 -0500 (EST) Cc: freebsd-current@freefall.cdrom.com In-Reply-To: <199503050749.JAA08984@smile.clinet.fi> from "Heikki Suonsivu" at Mar 5, 95 09:49:00 am X-Mailer: ELM [version 2.4 PL24] Content-Type: text Content-Length: 1731 Sender: current-owner@FreeBSD.org Precedence: bulk Heikki Suonsivu writes: > > > # Beginning with FreeBSD 2.1 you can wire down your SCSI devices so > # that a given bus, target, and LUN always come on line as the same > # device unit. In earlier versions the unit numbers were assigned > # in the order that the devices were probed on the SCSI bus. This > # means that if you removed a disk drive, you may have had to rewrite > # your /etc/fstab file, and also that you had to be careful when adding > # a new disk as it may have been probed earlier and moved your device > # configuration around. > > # This old behavior is maintained as the default behavior. The unit > # assignment begins with the first non-wired down unit for a device > # type. For example, if you wire a disk as "sd3" then the first > # non-wired disk will be assigned sd4. > > Would it be better to by default try to wire the first 4 disks? I'm not sure what you mean. Where would we put the disks on my system: bus 0 target 0 lun 0 bus 0 target 1 lun 0 bus 0 target 3 lun 0 Right now the behavior is easy to change (change scsi_assign_unit() in scsiconf.c) but after 2.1 it will be there for life, so let us be sure we like the approach. I don't think it is practical to do anything that changes the default behavior present in 2.0 and earlier. With what I've put in, all old kernels are supposed to boot and mount the drives in the same place they always have, with the same dangers we've always had. I shudder to think what the last week would have been like if I had changed the default behavior for disk assignments. -- Peter Dufault Real Time Machine Control and Simulation HD Associates, Inc. Voice: 508 433 6936 dufault@hda.com Fax: 508 433 5267