From owner-freebsd-current@FreeBSD.ORG Thu Dec 25 09:13:32 2003 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id D2DD116A4CE; Thu, 25 Dec 2003 09:13:32 -0800 (PST) Received: from harmony.village.org (rover.bsdimp.com [204.144.255.66]) by mx1.FreeBSD.org (Postfix) with ESMTP id C0A8943D1D; Thu, 25 Dec 2003 09:13:29 -0800 (PST) (envelope-from imp@bsdimp.com) Received: from localhost (warner@rover2.village.org [10.0.0.1]) by harmony.village.org (8.12.10/8.12.9) with ESMTP id hBPHDPAS073988; Thu, 25 Dec 2003 10:13:26 -0700 (MST) (envelope-from imp@bsdimp.com) Date: Thu, 25 Dec 2003 10:13:00 -0700 (MST) Message-Id: <20031225.101300.93473676.imp@bsdimp.com> To: ru@freebsd.org From: "M. Warner Losh" In-Reply-To: <20031225094520.GB23939@FreeBSD.org.ua> References: <20031222220014.GA659@mobi.lestinsky.de> <20031224.180046.83980462.imp@bsdimp.com> <20031225094520.GB23939@FreeBSD.org.ua> 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: michael@lestinsky.de cc: freebsd-current@freebsd.org Subject: Re: [RELENG_5_2] Cardbus trouble X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 25 Dec 2003 17:13:33 -0000 In message: <20031225094520.GB23939@FreeBSD.org.ua> Ruslan Ermilov writes: : On Wed, Dec 24, 2003 at 06:00:46PM -0700, M. Warner Losh wrote: : [...] : > You may try setting hw.cbb.start_memory as well. : > : Could you remind me, what's the story behind it? My cards : are only detected when I set it in /boot/loader.conf, per : your suggestion, but why is it necessary at all. I know : that PCI bus code lacks some support for it, but this is : more of a question why it is neccessary to _change_ the : default value of this variable on some machines? Because the current pci bus code assumes that the BIOS will assign all resources and is ill prepared to do anything beyond the most rudamentary assignments itself. It also bogusly doesn't allocate the resources it sees on device enumeration (*ALL* decoded resources should be allocated, even if the driver doesn't wind up using them). I've got some code that starts to fix these issues, but more work is needed. One of the biggest areas that needs work is to more fully support the heirarchical structure of the resources on the pci bus. Warner