On 03/19/2014 01:01 PM, Mark Brown wrote:
On Wed, Mar 19, 2014 at 12:57:57PM +0100, Lars-Peter Clausen wrote:
On 03/19/2014 12:53 PM, Mark Brown wrote:
Like I've said before it is not obvious to me that removing the wrappers is a win - yes, things can obtain their regmap but if enough things need to do it then having a convenience wrapper which does the lookup at least doesn't hurt.
It's more about things like snd_soc_update_bits. The wrappers will stay, but can as you said be made static inline functions and moved to the header.
I'm confused - that's what I said isn't it?
I think we are on the same page in regards to were we want to go.
I never said that I wanted to remove the wrappers, they hide the function from the implementation which is good in two ways, the user doesn't have to deal with the details implementation and secondly we can update the implementation without having to update the users.
What this series does, first moves everything related to ASoC level IO to soc-io.c and then starts cleaning it up. For now we still have to maintain some of the ASoC level IO implementation like snd_soc_update_bits() since not all drivers are using regmap yet. Once all have been converted snd_soc_component_write(...) boils down to regmap_write(component->regmap, ...), then we can move it to a header and make it static inline and remove soc-io.c.