From owner-freebsd-current Sun Apr 19 21:39:44 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id VAA16708 for freebsd-current-outgoing; Sun, 19 Apr 1998 21:39:44 -0700 (PDT) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: from antipodes.cdrom.com (castles147.castles.com [208.214.165.147]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id EAA16661 for ; Mon, 20 Apr 1998 04:39:40 GMT (envelope-from mike@antipodes.cdrom.com) Received: from antipodes.cdrom.com (localhost [127.0.0.1]) by antipodes.cdrom.com (8.8.8/8.8.5) with ESMTP id VAA03937; Sun, 19 Apr 1998 21:36:53 -0700 (PDT) Message-Id: <199804200436.VAA03937@antipodes.cdrom.com> X-Mailer: exmh version 2.0zeta 7/24/97 To: "Matthew N. Dodd" cc: Chuck Robey , current@FreeBSD.ORG Subject: Re: source code layout? In-reply-to: Your message of "Sun, 19 Apr 1998 22:36:38 EDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 19 Apr 1998 21:36:50 -0700 From: Mike Smith Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > On Sun, 19 Apr 1998, Chuck Robey wrote: > > On Sun, 19 Apr 1998, Matthew N. Dodd wrote: > > > Ok, we've got sys/pci for PCI devices which as near as I can tell is > > > fairly i386 specific > > > > Is that true? But, AFAIK, all the Alpha machines I've found have PCI in > > them also. I don't know enough to be really sure, but I'm skeptical of > > that claim. It might be that the pci stuff is currently i386 specific > > only because out alpha port hasn't gone far enough. > > AFAIK anywhere you see inb/outb/inw/outw you're fairly certain to be i386 > specific. (This from listening to the netbsd lits.) That's correct. However the use of in*/out* in sys/pci is a consequence of our lack of bus space support (to use the NetBSD term). We seem to have a general consensus that this is the right way to go about it, and part of the CAM work involves supplying infrastructure for this. > We've got sys/i386/scsi with very little in it, no sys/i386/net and a lot > of junk in sys/i386/isa/ic which should be in sys/ic or sys/dev/ic or > something. There should be *less* architecture-related code, not more. There is almost nothing that I can think of that should be in i386/net. In your specific case, you have a card architecture that can be accessed via several different busses. The general technique for dealing with this is exemplified in many drivers - the bulk of the driver lives in the source file associated with one of the busses (often isa), and the glue for other busses is moved accordingly. There has been some resistance to the use of sys/dev for the parking of general drivers, but IMHO if people were to start using more descriptive names there, it would be more suitable than anywhere else for bus- and architecture-neutral driver source. > I suppose I should steel myself and just ignore all of this as I've no > time to task against doing anything towards resolving my complaints. Unless you're willing to take on the enormous work involved in reorganising, and fight the battles over losing the CVS history, yes, there are better things to do. 8) -- \\ Sometimes you're ahead, \\ Mike Smith \\ sometimes you're behind. \\ mike@smith.net.au \\ The race is long, and in the \\ msmith@freebsd.org \\ end it's only with yourself. \\ msmith@cdrom.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message