From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756386AbZE1DGK (ORCPT ); Wed, 27 May 2009 23:06:10 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754975AbZE1DFq (ORCPT ); Wed, 27 May 2009 23:05:46 -0400 Received: from bilbo.ozlabs.org ([203.10.76.25]:39475 "EHLO bilbo.ozlabs.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751894AbZE1DFp (ORCPT ); Wed, 27 May 2009 23:05:45 -0400 Date: Thu, 28 May 2009 12:57:18 +1000 From: David Gibson To: Robert Schwebel Cc: Scott Wood , devicetree-discuss , linux-kernel@vger.kernel.org, Timur Tabi , Janboe Ye , linux-arm-kernel@lists.arm.linux.org.uk, rmk@arm.linux.org.uk Subject: Re: [RFC] [PATCH] Device Tree on ARM platform Message-ID: <20090528025718.GC1464@yookeroo.seuss> Mail-Followup-To: Robert Schwebel , Scott Wood , devicetree-discuss , linux-kernel@vger.kernel.org, Timur Tabi , Janboe Ye , linux-arm-kernel@lists.arm.linux.org.uk, rmk@arm.linux.org.uk References: <1243408083.13460.14.camel@debian-nb> <20090527150527.GK6805@pengutronix.de> <20090527162000.GM6805@pengutronix.de> <20090527234801.GP6805@pengutronix.de> <4A1DD3A0.3010501@freescale.com> <20090528000707.GR6805@pengutronix.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20090528000707.GR6805@pengutronix.de> User-Agent: Mutt/1.5.18 (2008-05-17) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, May 28, 2009 at 02:07:07AM +0200, Robert Schwebel wrote: > On Wed, May 27, 2009 at 06:58:24PM -0500, Scott Wood wrote: > > Robert Schwebel wrote: > >> The oftree by design wants to be a complete hardware description. As > >> you mention above, there are cases where you *nevertheless* need > >> ad-hoc information about things *not* encoded into the device tree. > >> > >> This renders the whole concept ad absurdum. You need a machine number > >> again - and if you need that: why not stay with the ARM model, define > >> everything with platform data and avoid the whole thing? > > > > Because it's better to have a little platform specific code than a lot > > of it? > > Until now, oftree has created more problems than it has solved for us. > The idea works fine for well-known things like memory maps and > interrupts. It works badly for corner cases, and embedded land is full > of it. The effort to get the oftree stuff right is often more than a > magnitude of order higher than the effort for the actual functionality. > That should be an alarm sign that something is wrong. It's true that going to a device tree model does take quite a bit of time and effort to get up to speed. As Grant said, we've made quite a few errors along the way on powerpc, as people experimented with the unfamiliar idea. Along the way a lot of people complained about how we were much better off without the device tree. However, I think nowadays, with most of the community at least roughly up to speed on device tree conventions, and the worst mistakes over, a lot of people who doubted the value of the device tree approach are coming around. Grant mentions the Xilinx devices as an example - they're interesting because of the fancy stuff they're doing, autogenerating the devtree from the FPGA configuration. But to me the real indicator of the device tree's value is the handful of platforms for which support has been added with *only* a new dts - a single, fairly simple text file. -- David Gibson | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_ | _way_ _around_! http://www.ozlabs.org/~dgibson From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Gibson Subject: Re: [RFC] [PATCH] Device Tree on ARM platform Date: Thu, 28 May 2009 12:57:18 +1000 Message-ID: <20090528025718.GC1464@yookeroo.seuss> References: <1243408083.13460.14.camel@debian-nb> <20090527150527.GK6805@pengutronix.de> <20090527162000.GM6805@pengutronix.de> <20090527234801.GP6805@pengutronix.de> <4A1DD3A0.3010501@freescale.com> <20090528000707.GR6805@pengutronix.de> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Content-Disposition: inline In-Reply-To: <20090528000707.GR6805-bIcnvbaLZ9MEGnE8C9+IrQ@public.gmane.org> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: devicetree-discuss-bounces+gldd-devicetree-discuss=m.gmane.org-mnsaURCQ41sdnm+yROfE0A@public.gmane.org Errors-To: devicetree-discuss-bounces+gldd-devicetree-discuss=m.gmane.org-mnsaURCQ41sdnm+yROfE0A@public.gmane.org To: Robert Schwebel Cc: devicetree-discuss , linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Timur Tabi , Scott Wood , Janboe Ye , linux-arm-kernel-xIg/pKzrS19vn6HldHNs0ANdhmdF6hFW@public.gmane.org, rmk-lFZ/pmaqli7XmaaqVzeoHQ@public.gmane.org List-Id: devicetree@vger.kernel.org On Thu, May 28, 2009 at 02:07:07AM +0200, Robert Schwebel wrote: > On Wed, May 27, 2009 at 06:58:24PM -0500, Scott Wood wrote: > > Robert Schwebel wrote: > >> The oftree by design wants to be a complete hardware description. As > >> you mention above, there are cases where you *nevertheless* need > >> ad-hoc information about things *not* encoded into the device tree. > >> > >> This renders the whole concept ad absurdum. You need a machine number > >> again - and if you need that: why not stay with the ARM model, define > >> everything with platform data and avoid the whole thing? > > > > Because it's better to have a little platform specific code than a lot > > of it? > > Until now, oftree has created more problems than it has solved for us. > The idea works fine for well-known things like memory maps and > interrupts. It works badly for corner cases, and embedded land is full > of it. The effort to get the oftree stuff right is often more than a > magnitude of order higher than the effort for the actual functionality. > That should be an alarm sign that something is wrong. It's true that going to a device tree model does take quite a bit of time and effort to get up to speed. As Grant said, we've made quite a few errors along the way on powerpc, as people experimented with the unfamiliar idea. Along the way a lot of people complained about how we were much better off without the device tree. However, I think nowadays, with most of the community at least roughly up to speed on device tree conventions, and the worst mistakes over, a lot of people who doubted the value of the device tree approach are coming around. Grant mentions the Xilinx devices as an example - they're interesting because of the fancy stuff they're doing, autogenerating the devtree from the FPGA configuration. But to me the real indicator of the device tree's value is the handful of platforms for which support has been added with *only* a new dts - a single, fairly simple text file. -- David Gibson | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_ | _way_ _around_! http://www.ozlabs.org/~dgibson