From owner-freebsd-hackers Fri May 12 10:21:21 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id KAA13838 for hackers-outgoing; Fri, 12 May 1995 10:21:21 -0700 Received: from cs.weber.edu (cs.weber.edu [137.190.16.16]) by freefall.cdrom.com (8.6.10/8.6.6) with SMTP id KAA13832 for ; Fri, 12 May 1995 10:21:19 -0700 Received: by cs.weber.edu (4.1/SMI-4.1.1) id AA01377; Fri, 12 May 95 11:13:14 MDT From: terry@cs.weber.edu (Terry Lambert) Message-Id: <9505121713.AA01377@cs.weber.edu> Subject: Re: No PS/2 mouse driver in default kernel To: rgrimes@gndrsh.aac.dev.com (Rodney W. Grimes) Date: Fri, 12 May 95 11:13:13 MDT Cc: jkh@freefall.cdrom.com, babkin@hq.icb.chel.su, hackers@FreeBSD.org In-Reply-To: <199505120415.VAA14085@gndrsh.aac.dev.com> from "Rodney W. Grimes" at May 11, 95 09:15:55 pm X-Mailer: ELM [version 2.4dev PL52] Sender: hackers-owner@FreeBSD.org Precedence: bulk > Not quite right, Terry is going to jump in here I am sure, is what > needs to happen is the keyboard code needs to be ripped out of > both syscons and pcvt and put in a separate driver. This new driver > needs to handle the PS/2 mouse, and the keyboard, but nothing else. Right you are. And the PS/2 mouse driver would never be accessed directly; there would be translation to a common mouse protocol and it would be fed up through /dev/mouse (as would all the other mouse drivers). > This change would reduce a large chunk of duplicate code between > pcvt, syscons and psm. Amen. > There was tons of architectural discussion on what was once the console > driver mailing list. I am sure Terry has this all rat holed some > place on his system. I do. I also have all the subsequent discussions regarding the console in any way, shape or form. Actually, I've "rat-holed" most of the other major subsystem discussions I've ever participated in (like all the file system issues for every file system type). I even have most of the initial Bill Jolitz shared library discussions squirreled away (though stuff that old is on tape). I'm sure most of this is on the hackers or arch or news archives anyway. > That group of folks did come up with a very good layered architecture > that should be used to fix this whole mess. It did. I disagreed with some points, but then I always have been a maniacal purist when it comes to making design tradeoffs. I count my instructions and I point out that a while loop uses jump's in the assembly code just like goto's (anti-goto freaks don't understand the connection between source code and assembly code, and they don't get the ideas behing single entry/exit and exception handling). I guess it comes from growing up on a 4k machine. Color me rabid. 8-). (I can already hear the cries of "4K?!? >*LUXURY**