From owner-freebsd-hackers Sat Feb 18 03:46:29 1995 Return-Path: hackers-owner Received: (from root@localhost) by freefall.cdrom.com (8.6.9/8.6.6) id DAA12683 for hackers-outgoing; Sat, 18 Feb 1995 03:46:29 -0800 Received: from hda.com (hda.com [199.232.40.182]) by freefall.cdrom.com (8.6.9/8.6.6) with ESMTP id DAA12677 for ; Sat, 18 Feb 1995 03:46:24 -0800 Received: (dufault@localhost) by hda.com (8.6.9/8.3) id GAA26730; Sat, 18 Feb 1995 06:44:48 -0500 From: Peter Dufault Message-Id: <199502181144.GAA26730@hda.com> Subject: Re: NEW_SCSICONF To: julian@tfs.com (Julian Elischer) Date: Sat, 18 Feb 1995 06:44:48 -0500 (EST) Cc: freebsd-hackers@freefall.cdrom.com In-Reply-To: from "Julian Elischer" at Feb 16, 95 11:26:26 pm X-Mailer: ELM [version 2.4 PL24] Content-Type: text Content-Length: 2606 Sender: hackers-owner@FreeBSD.org Precedence: bulk Julian Elischer writes: > > > > > Can someone explain the history of "NEW_SCSICONF", and whether it should > > be left in there? I'd like to remove it. > The NEW_SCSICONF stuff is an attempt to make some of the features > of teh tape driver (e.g. rogue entries) more general to the > SCSI system as a whole, and to rationalise some of the scsi > systems's configuration actions and interfaces. It was implimented > by the PCI driver crew, and has some good features. Don't just delete > it. How about keeping only NEW_SCSICONF and removing the default case? Is there any reason to keep them both? I'll turn on NEW_SCSICONF to make sure nothing breaks. Also regarding NEW_SCSICONF: how about pushing the broken tape drive data structures down into st and the broken disk drive data structures down into sd, etc, instead of that single large table in scsiconf.c? I bet a lot of the #defines could be known only in the "type" driver (st, st, etc) and we could reduce the exported stuff in scsiconf.h. Much of it is there to provide a coupling between scisconf and the "type" driver. > > > > > Also, is anyone using the "predef" stuff in scsiconf.c? I'd like to > > remove it and replace it with a traditional config approach. > yes I have used it but only on the OSF version of the code... I removed it during adding the config code. I haven't committed that yet, though. > > Some of the things that need doing.. > 1/ examination of charles's changes in NetBSD, and the adoption of some > quite astute simplifications and reallignments he's done.. (also bug-fixes) > The difficult part in this is that he''s reformatted everything again, > and the NetBSD config stuff is considerably different, leading to quite > different softc structures and configuration routines. Can you say what some of his astute simplifications are? Where is the NetBSD code archived on freefall? > 2/ examination of the NEW_SCSICONF code and extraction of pertinant ideas > and changes. Or making NEW_SCSICONF the only SCSICONF, and possibly pushing it down into the type drivers as I mentioned before. > 3/ both sd and wd drivers should share more "how a disk behaves" code.. > 4/ the tape driver needs a good shake-out > 5/ the cd driver needs code for audio-read (I have that for toshiba drives > at least). possibly dependent on drive manufacturer. > 6/ all-round documentation of interfaces... comments checked for correctness. Peter -- Peter Dufault Real Time Machine Control and Simulation HD Associates, Inc. Voice: 508 433 6936 dufault@hda.com Fax: 508 433 5267