On Wed, Aug 01, 2018 at 12:47:01PM +0200, Alexander Sverdlin wrote:
> On 01/08/18 12:39, Mark Brown wrote:

> > I'm not 100% clear I follow what you mean by bus scan here but I *think*
> > that sounds about right.  The channel request/release could be factored
> > out into helper functions to minimize duplication.

> Rob has an opinion that the driver should only claim the channels explicitly
> specified in DT. This sounds about right, but this would be behavioral change.
> What do you think?

If it results in platforms that previously used DMA not using DMA that
doesn't seem like a good idea.  I'd have thought that if we didn't have
DMA configured we'd already not be able to request a channel TBH.