On 08/07/11 14:08, Marc Zyngier wrote:
> So you're basically folding of_core_device_populate() and
> core_driver_init_class() into one call, and generating the
> of_device_ids on the fly. If you're going down that road,
> it would be even simpler to directly use of_device_ids
> instead of core_device_ids and skip the generation altogether.
> 
> That would also remove the static declaration of devices to be
> probed in the architecture support code...
> 
> Let me think of it and prototype that.

See the attached patch against branch dt_gic_twd from
git://git.kernel.org/pub/scm/linux/kernel/git/maz/arm-platforms.git

It boots fine on my PB11MP.

What do you think?

	M.
-- 
Jazz is not dead. It just smells funny...