From owner-freebsd-arch@FreeBSD.ORG Mon Aug 2 01:27:19 2004 Return-Path: Delivered-To: freebsd-arch@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id C0D4C16A4CE for ; Mon, 2 Aug 2004 01:27:19 +0000 (GMT) Received: from harmony.village.org (rover.village.org [168.103.84.182]) by mx1.FreeBSD.org (Postfix) with ESMTP id 35CED43D4C for ; Mon, 2 Aug 2004 01:27:19 +0000 (GMT) (envelope-from imp@bsdimp.com) Received: from localhost (warner@rover2.village.org [10.0.0.1]) by harmony.village.org (8.12.11/8.12.11) with ESMTP id i721OYs9001230; Sun, 1 Aug 2004 19:24:39 -0600 (MDT) (envelope-from imp@bsdimp.com) Date: Sun, 01 Aug 2004 19:25:04 -0600 (MDT) Message-Id: <20040801.192504.98673359.imp@bsdimp.com> To: scottl@samsco.org From: "M. Warner Losh" In-Reply-To: <410D5F45.1060902@samsco.org> References: <61725.1091393254@critter.freebsd.dk> <410D5F45.1060902@samsco.org> X-Mailer: Mew version 3.3 on Emacs 21.3 / Mule 5.0 (SAKAKI) Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit cc: arch@freebsd.org cc: phk@phk.freebsd.dk Subject: Re: PCI-Express support X-BeenThere: freebsd-arch@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussion related to FreeBSD architecture List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 02 Aug 2004 01:27:19 -0000 In message: <410D5F45.1060902@samsco.org> Scott Long writes: : Agreed, and I'm looking for things besides stability that are going to : bite us in the near future. 6.0 likely won't happen for 12-months : after 5.3 at an absolute minimum, and I don't want PCI-E support to : turn into something like Cardbus is with 4.x vs 5.x. A big part of why CardBus evolved the way it did was because it depended on pci extensions that Mike Smith explicitly said he wasn't going to back port. Later, it was because of the acpi support in current that wasn't in stable (most modern laptops do not operate correctly with cardbus cards if you don't use acpi). Since the changes were extensive and all over the tree, it was hard to back port. Also, there was much optimism about the possibility of a quick 5.0, which evolved the way we all know. I think that we should learn from this history that we should be less afraid to back port when there's a highly anticipated feature. We should be as pessimistic as possible when guessing the date for 6.0 for the purposes of deciding if something should go into the tree. Eg, if we think that we can do 6.0 on a certain date, we should add a year or two to that when deciding to back port or not. Had we applied that rule with the CardBus code, we'd have back ported things w/o hesitation. In fact, there was a preliminary port to -stable at the time we branched, and that just wasn't kept up to date. We started doing major changes too quickly, which made it harder to back port. Anyway, I just wanted to speak up from my perspective from CardBus. Warner