* Tony Lindgren <tony@atomide.com> [080916 16:14]:
> * Tony Lindgren <tony@atomide.com> [080913 10:12]:
> > * Russell King - ARM Linux <linux@arm.linux.org.uk> [080913 03:00]:
> > > On Thu, Sep 11, 2008 at 05:48:49PM -0700, Tony Lindgren wrote:
> 
> <snip>
> 
> > > Maybe these functions should also be taking note of info->nr_slots?
> > > Though I don't particularly like the way 'info' is shared between both
> > > controllers.  It's more usual to pass a data structure to drivers
> > > describing just the data for _this_ instance of the device.
> > 
> > Let's see if that can be sorted out too. Some implementations multiplex
> > the slot over I2C.
> > 
> > > Now, when you come across a device with three controllers, you're not
> > > modifying arch/arm/plat-omap/devices.c to add that third controller -
> > > you're just creating the omapN_init_mmc() function with another
> > > omap_mmc_add() line.
> > > 
> > > As an added bonus, notice the lack of nasty #ifdef's scattered in there
> > > as well.
> > 
> > Yeah that's good.
> > 
> > > Oh, and I see little reason for checking for NULL data in these functions -
> > > they clearly don't take NULL data, and if someone passes NULL data, they
> > > deserve to oops - and hopefully they'll fix their code.
> > 
> > OK
> 
> Here's the patch updated. If necessary, the patch could be broken into
> several parts.
> 
> Turns out that the omap1 MMC support has been pretty badly broken for
> a while now. I ended up removing all the obviously broken MMC
> configurations, and only try to revive the ones that seem like
> they should work. I also removed the cover switch handling that needs
> to be reimplemented using gpio chip instead.

Here's the patch with few compile and whitespace warnings fixed.

> Regards,
> 
> Tony