From owner-freebsd-alpha Fri Jan 9 18:33:06 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.7/8.8.7) id SAA10502 for alpha-outgoing; Fri, 9 Jan 1998 18:33:06 -0800 (PST) (envelope-from owner-freebsd-alpha@FreeBSD.ORG) Received: from implode.root.com (implode.root.com [198.145.90.17]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id SAA10497 for ; Fri, 9 Jan 1998 18:32:54 -0800 (PST) (envelope-from root@implode.root.com) Received: from implode.root.com (localhost [127.0.0.1]) by implode.root.com (8.8.5/8.8.5) with ESMTP id SAA12647; Fri, 9 Jan 1998 18:31:30 -0800 (PST) Message-Id: <199801100231.SAA12647@implode.root.com> To: "Justin T. Gibbs" cc: Curt Sampson , alpha@FreeBSD.ORG Subject: Re: Alpha port.. In-reply-to: Your message of "Fri, 09 Jan 1998 18:09:24 MST." <199801100109.SAA07399@narnia.plutotech.com> From: David Greenman Reply-To: dg@root.com Date: Fri, 09 Jan 1998 18:31:30 -0800 Sender: owner-freebsd-alpha@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org >> As far as this actually happening on the FreeBSD side, I'd love to >> see a lot of this stuff adopted into FreeBSD, since we rely on a >> couple of FreeBSD folks for some of our device drivers. However, >> last time I chatted with David Greenman (at ISPCon in SF last >> August) he indicated to me that he felt that inb() and outb() were >> just fine for device drivers, and he saw no need for any of the >> bus_space stuff in NetBSD. > >If that was indeed David's position at the time, I think it's changed. >All of bus.h as well as enough of the Bus DMA stuff to do page based >bounce buffers has already been implemented and is available in the >CAM SCSI layer snapshots. I should probably push to get these changes >formally reviewed an put into the tree though. They aren't really CAM >specific. Yeah, that wouldn't be an entirely accurate quote. What I recall saying was that 1) FreeBSD had no plans to support a large number of platforms like NetBSD does (many reasons that I'll not go into now), 2) the bus.h stuff seemed more wiz-bang complicated than I thought it needed to be for the platforms we'd likely support, 3) until we started supporting another architecture I didn't see the need to integrate the NetBSD stuff - we have many other more important things to worry about. I think I also said that it simple isn't the focus of the FreeBSD Project to be completely architecture neutral in all of our device drivers, especially when it means making performance tradeoffs to accomplish it. I realize that this is a fundamental difference in the focus and direction of the two projects and is a major reason why we've never been able to merge. Now, Justin wanted to support NetBSD in his Adaptec device driver and wanted to take on porting the NetBSD bus.h implementation as well as improve upon it. I have no problem with that (and yes, Justin and I discussed the issue at one point) and I applaud his efforts. Does this make my position any more clear? It really comes down to priorities, and until Justin needed it for easier NetBSD support, it was not a priority of anyone working on FreeBSD. -DG David Greenman Core-team/Principal Architect, The FreeBSD Project