On Wed, Nov 23, 2011 at 11:33:53AM +0530, Vinod Koul wrote:
On Tue, 2011-11-22 at 19:43 +0000, Mark Brown wrote:
Looking at those they're going to need the regulator bindings for device tree to do anything useful. I'd expect that to go into 3.3 but it'll be hassle propagating the binding bits.
I have a question here, how does using DT to a codec (which can be used even on a system where there is no DT present) be generic and break the assumption that codec does not know anything about the system it operates in? This should be done for machine or cpu driver which know if system has DT or not?
When drivers implement device tree support this should be done in addition to rather than instead of platform data support so other platforms should not be affected. A very common way of implementing this is that the device tree parsing code translates the device tree into a platform data structure and then the driver runs off a copy of the platform data structure.