On Tue, Jan 18, 2022 at 11:21:50AM -0600, Pierre-Louis Bossart wrote:
On 1/18/22 11:10 AM, Mark Brown wrote:
Why is this a good fix and not removing the dependency on gpiolib from DMIC? While a DMIC *can* use a GPIO it's not something that's an intrinsic requirement and it's entirely optional in the code.
We also have similar 'depends on GPIOLIB' for Intel boards that are not really useful. see e.g. 4262ddc2ad63 ('ASoC: Intel: boards: add explicit dependency on GPIOLIB when DMIC is used'). IIRC we had this discussion before about doing a larger cleanup.
At first glance it's just that the gpiolib dependency on the DMIC driver is spurious and should be deleted, the code looks fine and I'd not expect any need for an actual dependency (but I'm out of office this week so probably not going to actually poke at things properly right now).