[RFC PATCH 0/3] Separate BE DAI HW constraints from FE ones

[Codrin.Ciubotariu at microchip.com]

On 15.04.2021 20:25, Mark Brown wrote:
> On Thu, Apr 15, 2021 at 04:56:00PM +0000, Codrin.Ciubotariu at microchip.com wrote:
> 
>> Are there any plans for refactoring DPCM? any ideas ongoing? I also have
>> some changes for PCM dmaengine, in the same 'style', similar to what I
>> sent some time ago...
>> I can adjust to different ideas, if there are any, but, for a start, can
>> anyone confirm that the problem I am trying to fix is real?
> 
> Lars-Peter's presentation from ELC in 2016 (!) is probably the clearest
> summary of the ideas:
> 
>     https://elinux.org/images/e/e7/Audio_on_Linux.pdf
>     https://youtu.be/6oQF2TzCYtQ
> 
> Essentially the idea is to represent everything, including the DSPs, as
> ASoC components and be able to represent the digital links between
> components in the DAPM graph in the same way we currently represent
> analogue links.  This means we need a way to propagate digital
> configuration along the DAPM graph (or a parallel, cross linked digital
> one).  Sadly I'm not really aware of anyone actively working on the
> actual conversion at the minute, Morimoto-san has done a lot of great
> preparatory work to make everything a component which makes the whole
> thing more tractable.
> 

Thank you for the links! So basically the machine driver disappears and 
all the components will be visible in user-space.
If there is a list with the 'steps' or tasks to achieve this? I can try 
to pitch in.

Best regards,
Codrin