From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Thu, 14 Jun 2007 14:29:07 +1000 From: David Gibson To: Segher Boessenkool Subject: Re: [PATCH] powerpc: Create "rom" (MTD) device prpmc2800 Message-ID: <20070614042906.GC11177@localhost.localdomain> References: <7878cf1aec340b976b90b86b9e83bf18@kernel.crashing.org> <20070612044246.GC4198@localhost.localdomain> <9fbd7a7f5cdde58768569ab23c7aec7c@kernel.crashing.org> <20070613061254.GG16148@localhost.localdomain> <20070613091904.GA30948@localhost.localdomain> <791ab8b82c5b5d2b3ae4827a891a43ea@kernel.crashing.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <791ab8b82c5b5d2b3ae4827a891a43ea@kernel.crashing.org> Cc: linuxppc-dev List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Wed, Jun 13, 2007 at 11:37:42AM +0200, Segher Boessenkool wrote: > >> How do non-device-tree MTD users handle this? > > > > Invidual, board-specific map files with hardcoded partition tables. > > That's what I was afraid of. Sigh. > > > See drivers/mtd/maps/ebony.c for example. > > [/me looks...] Oh horror. > > > Hence my view of the device > > tree as a preferable option. > > I think some binary partition table format that can be > used by _all_ MTD consumers should be defined. How > that table should be communicated to the kernel in the > device tree case we can discuss later, then. Maybe > something as simple as storing it in flash, and have a > "partition-table-offset" property or something like that. > > This is something the MTD people will have to buy into > of course. I tought I saw some config option implying that there already existed an on-device partition table format for flashes. Doesn't help us for existing boards which don't expect such a setup, of course. Incidentally, with either the partitions-described-by-properties, or the revised more-flexible partitions-described-by-subnodes format, I'm not seeing it as required that the whole of the flash address space be described. So it would certainly be possible to *only* describe the sections of the flash used by firmware. But I'm suggesting that we optionally allow other partitions to be described for boards / systems which have strong conventions about how the flash is divided and where we don't have any other way of recording the partition layout short of hardcoding. -- 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