From owner-freebsd-hackers Mon May 8 09:22:52 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id JAA03806 for hackers-outgoing; Mon, 8 May 1995 09:22:52 -0700 Received: from kitten.mcs.com (Kitten.mcs.com [192.160.127.90]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id JAA03798 ; Mon, 8 May 1995 09:22:48 -0700 Received: from mailbox.mcs.com (Mailbox.mcs.com [192.160.127.87]) by kitten.mcs.com (8.6.10/8.6.9) with SMTP id LAA14789; Mon, 8 May 1995 11:22:00 -0500 Received: by mailbox.mcs.com (/\==/\ Smail3.1.28.1 #28.5) id ; Mon, 8 May 95 11:12 CDT Received: by mercury.mcs.com (/\==/\ Smail3.1.28.1 #28.5) id ; Mon, 8 May 95 11:12 CDT Message-Id: Subject: Re: SLICES and bits in the device numbers To: julian@ref.tfs.com (Julian Elischer) Date: Mon, 8 May 1995 11:12:41 -0500 (CDT) From: "Lars Fredriksen" Cc: dufault@hda.com, bde@zeta.org.au, hackers@FreeBSD.org, julian@FreeBSD.org In-Reply-To: <199505060028.RAA04681@ref.tfs.com> from "Julian Elischer" at May 5, 95 05:28:13 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1586 Sender: hackers-owner@FreeBSD.org Precedence: bulk Julian Elischer writes: > > > > > Couldn't we use the "bus" device for all lowlevel stuff like formatting, > > tuning, tape fixing etc? What I mean is that we open /dev/scsi0 and > > blast the scsi request straight in that way. It does mean some code > > duplication between the different drivers and the "control" program, > > but if we are running out of bits, it might be worth it. > but you couldn't blast a format at sd0 but at sb0t0l0 or whatever, > which might be easier to get wrong? Forgot about formatting the partitions. Yes you are right, its going to have to be per device. Even though, I still would like to see us be able to write directly to the different SCSI devices without going through the different dirvers. There seems to be to much stuff hapening on open/close when we go through existing device drivers to get to the SCSI device. > > my own suggestion is that slice1 (presently whole disk slice) be divided > into a couple of sub-devices.. > 0: whole disk > 1: MBR block (if it exists) > 2: control (ioctl only) device > > the st driver already has a control device.. > What a control device actually means is dependent on the kind of device, > (setting modes permanently on tapes is one example) > so I see no reason to not make it's definition device specific, > unless you want to make teh SCSI code define this.. > (which may not be what we want) > > julian > -- ------------------------------------------------------------------- Lars Fredriksen fredriks@mcs.com (home) lars@fredriks.pr.mcs.net (home-home) fredriks@asiago.cs.wisc.edu