Re: [patch/rfc 2.6.29 1/2] MTD: driver model updates

[David Woodhouse <dwmw2@infradead.org>]

On Fri, 2009-04-03 at 13:00 -0700, Kevin Cernekee wrote:
> Based on: http://lists.infradead.org/pipermail/linux-mtd/2009-March/025005.html
> 
> My only change from the previous posting
> (http://lists.infradead.org/pipermail/linux-mtd/2009-April/025121.html)
> was to remove the "mtd_" prefix on the device attributes.
> 
> David's 2/2 patch
> (http://lists.infradead.org/pipermail/linux-mtd/2009-March/025011.html)
> may still be used as-is.
> 
> Signed-off-by: Kevin Cernekee <kpc.mtd@gmail.com>

Thanks, this looks like a very good start in the direction we need to
go. I've applied David's patches as well as the changes from this one,
and hooked it up for the CAFÉ NAND controller too.

Since the callers are passing a struct mtd_info * into nand_scan(), it
doesn't seem necessary to pass the device in too; they can just set it
for themselves. Passing it in to the NOR chip probe routines might make
sense though.

I'm not worried about a flag day for _internal_ stuff -- for 2.6.31 I
think I'm going to add a WARN_ON(&mtd->dev.parent) into the core code.

I really do think I want this to avoid the need for 64-bit ioctls
(except maybe MEMERASE64).

-- 
David Woodhouse                            Open Source Technology Centre
David.Woodhouse@intel.com                              Intel Corporation