[Ksummit-2012-discuss] [ATTEND] Complex dependencies in device model

Arnd Bergmann arnd at arndb.de
Wed Jun 20 20:53:58 UTC 2012


On Wednesday 20 June 2012, Matthew Garrett wrote:
> On Wed, Jun 20, 2012 at 10:54:56AM -0600, Stephen Warren wrote:
> 
> > With all the recent focus on ARM cleanup and reviews, I hope we can
> > catch this kind of thing, so such restrictions wouldn't be necessary.
> > 
> > Not merging the architecture support unless the drivers are merged first
> > somewhat creates a chicken/egg problem; should the drivers be merged
> > without any architecture to run on? I can easily see someone else
> > wanting that restriction...
> 
> The issue is that FDT and ACPI are two mechanisms for providing the same 
> data. If we have 64-bit ARM hardware that can be configurd with both 
> then we'll end up with two sets of drivers - one using the FDT bindings, 
> one using the ACPI bindings. We should consolidate those before landing 
> an architecture that encourages this kind of divergence.

Most of the device drivers are shared with 32 bit ARM, which currently
has FDT bindings but no ACPI bindings. There is no way around using those
devices in FDT mode with the same bindings we use on 32-bit ARM or other
architectures that already share those drivers.

All the 64-bit ARM code that I've seen is FDT-only, and the drivers
are all shared, AFAICT, so there is no harm in supporting that.

What we do about ACPI is a completely different question that we
should discuss when someone submits ACPI code for ARM.

	Arnd


More information about the Ksummit-2012-discuss mailing list