Hey,

On Thu, 12 Mar 2020, Ville Syrjälä wrote:

> On Thu, Mar 12, 2020 at 07:27:58PM +0200, Kai Vehmanen wrote:
>> So I think this starts to look that we should move calling glk_force_audio 
>> to bind/unbind pair. I can make a patch for this.
> 
> That would stop us from doing dynamic cdclk changes once we get the hw
> that can do that properly. Rather I think I'd just hardcode the 2xbclk
> requirement in i915 for the platforms that suck.

well, you can always code in both places -- i.e. have the clock 
constraints set up at bind() for older (the current) platforms, and have a 
different callplace in get_power() for newer platforms. This code is 
anyways conditional on the hardware that is used.

I now send a patch implementing this, plus code to at runtime figure
out the effective BCLK, to the list. Please have a look at comment.

Br, Kai