On Thu, Mar 18, 2010 at 07:18, Mark Brown wrote:
On Thu, Mar 18, 2010 at 05:08:01PM +0800, Barry Song wrote:
On Thu, Mar 18, 2010 at 4:48 PM, Liam Girdwood wrote:
Btw, have a look at wm8731.c. This driver supports both I2C and SPI and is less churn than splitting out the code into three files.
We have many discussion about a driver to support both spi and i2c in another device(ad714x). At the beginning, the way in wm8731.c is used, then after many comments, people agree splitting the driver to three modules is the best way. If you are interested in it, you may take a look at "input/misc: add Analog Devices AD714x captouch input driver" in linux-input maillist.
This was a discussion for the input subsystem and what makes sense here. For ASoC we've been using the approach that WM8731 and other similar drivers are using, please follow that approach with ASoC drivers. Adding files per driver per bus type isn't helpful here.
that isnt really true. the tact taken in the input subsystem wrt multiple busses had nothing to do with the input subsystem. it was designed to provide flexibility -- the core is not tied to the busses, and any of the pieces could be loaded/unloaded on the fly instead of forcing the core to be rebuilt just because of bus changes.
i'm not suggesting the ASoC codecs shouldnt use the same style across the board (and thus we shouldnt change the AD193x driver), just pointing out your basic premise here is invalid and that there are pros/cons to each method and that they're independent of the subsystem. you simply selected a different solution. -mike