From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from moutng.kundenserver.de (moutng.kundenserver.de [212.227.126.186]) by ozlabs.org (Postfix) with ESMTP id DE90ADDEDF for ; Wed, 7 Feb 2007 21:44:39 +1100 (EST) From: Arnd Bergmann To: "Li Yang-r58472" Subject: Re: [PATCH 1/3] add USB setup code for 8349emds PB Date: Wed, 7 Feb 2007 11:44:29 +0100 References: <989B956029373F45A0B8AF02970818900D4339@zch01exm26.fsl.freescale.net> In-Reply-To: <989B956029373F45A0B8AF02970818900D4339@zch01exm26.fsl.freescale.net> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Message-Id: <200702071144.29548.arnd@arndb.de> Cc: linuxppc-dev@ozlabs.org, Paul List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Wednesday 07 February 2007 11:10, Li Yang-r58472 wrote: > How about a non-PCI ehci driver for all architectures (such as arm, m68k)? > Then, platform_bus is a more generic interface than of_platform. Yes, that's right. The main difference is that if you scan all of devices at boot time, you already have a node for the USB controller and you shouldn't need to create another node that does not reflect the actual hierarchy in the linux device tree. A platform_device is really meant to be 'something that can't be probed but we know it is there', like the interrupt controller on a PC. With the of device tree, we can probe anything, so ideally, we should not need to create any platform devices at all. Arnd <><