From owner-freebsd-chat@FreeBSD.ORG Thu Aug 31 22:57:08 2006 Return-Path: X-Original-To: freebsd-chat@freebsd.org Delivered-To: freebsd-chat@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id C0EC916A4E5; Thu, 31 Aug 2006 22:57:08 +0000 (UTC) (envelope-from miod@ribeyre.gentiane.org) Received: from ribeyre.gentiane.org (odyssee.gentiane.org [80.65.224.82]) by mx1.FreeBSD.org (Postfix) with ESMTP id 8070C43D6A; Thu, 31 Aug 2006 22:57:05 +0000 (GMT) (envelope-from miod@ribeyre.gentiane.org) Received: from ribeyre.gentiane.org (miod@localhost.gentiane.org [127.0.0.1]) by ribeyre.gentiane.org (8.13.6/8.13.4) with ESMTP id k7VMvLVQ024302; Thu, 31 Aug 2006 22:57:21 GMT Received: (from miod@localhost) by ribeyre.gentiane.org (8.13.6/8.13.4/Submit) id k7VMvK3Q023735; Thu, 31 Aug 2006 22:57:20 GMT Date: Thu, 31 Aug 2006 22:57:20 +0000 From: Miod Vallat To: "Marc G. Fournier" Message-ID: <20060831225719.GG25515@ribeyre.gentiane.org> References: <20060830232723.GU10101@multics.mit.edu> <98f5a8830608301731s2b0663e3g94b0bd32f8a06a78@mail.gmail.com> <950621ad0608310654h78ae0023g346abd108815ae72@mail.gmail.com> <20060831110112.J82634@hub.org> <20060831184715.B82634@hub.org> <44F7619B.8010609@evilkittens.org> <20060831192632.T82634@hub.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20060831192632.T82634@hub.org> User-Agent: Mutt/1.5.11 Cc: miros-discuss@mirbsd.org, misc@openbsd.org, netbsd-users@netbsd.org, Gilles Chehade , freebsd-chat@freebsd.org Subject: Re: The future of NetBSD X-BeenThere: freebsd-chat@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Non technical items related to the community List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 31 Aug 2006 22:57:08 -0000 > I'd rather have Adaptec provide a source code driver for their cards > directly, then have Scott Long have to fight with unavailability of > documentation itself ... if the driver works, what do we need > documentation for? To fix the driver. A given piece of source code can only been believed correct until it is proven to be broken (unless you're Knuth, but he did not write device drivers). Even if the driver API is static, the hardware you'll want to use your particular card on isn't. New hardware may (and will) need tweaks to work on your new Uberathlon128 system in two years. If the driver happens to be working by chance on x86 because of, say, some cache behaviour future (read: legacy-free) hardware won't guarantee, what are you going to do? Beg the vendor to fix the driver for this card while it wants you to buy the new, expensive, device flavour which sports 128-bit bells&whistles? Fixing a subtly broken piece of code might not be as simple as adding bus_dmamap_sync() calls here and there. Miod