[PATCH 6/6] ASoC: audio-graph: remove Platform support

[Olivier MOYSAN]

Hi Kuninori,

Thanks for your feedback

On 8/26/21 12:48 AM, Kuninori Morimoto wrote:
> 
> Hi Olivier
> 
> Thank you for conntacting me
> 
>> I have seen that the STM32MP15 audio sound card is no more functional
>> with recent kernels (5.13 or 5.14)
>> The sound card is registered, but the all devices are issuing an error
>> at runtime. These devices are using stm32_sai.c or stm32_i2s.c
>> drivers.
>>
>> I found that the regression is linked to the commit
>> 63f2f9cceb09f8e5f668e36c1cf764eea468ebed "ASoC: audio-graph: remove
>> Platform support", as reverting this commit fixes the issue.
>>
>> When the platform component is missing the pcm_construct ops in the
>> pcm dmaengine, is never called, resulting in an incomplete
>> initialization of the sound card.
>> I can't figure out what is the right way to handle this change, however.
>> Do I need to update the CPU drivers to work without a platform component
>> or does the audio-graph card has to be changed in some way ?
> 
> Ahh, OK, I see.
> Indeed the dev which is used for CPU is used at soc-generic-dmaengine as Platform,
> without indicating it at DT (= simple-card has "plat" support for platform at DT,
> but audio-graph doesn't ).
> 

Yes, it seems that there is no way to force CPU to be used as platform 
with audio-graph. so, asoc_simple_canonicalize_platform() is necessary 
to do the job in this case.

> I think key funciton is asoc_simple_canonicalize_platform().
> 
>> What do you mean "We can respawn it when we need it", in the commit
>> message ?
> 
> This means we can revert this patch if needed, and yes it is needed :)
> Could you please respawn the feature ? or I can do it if you want.
> 

I feel more confortable if you revert the commit, as you are the author 
of the patch.
Thanks.

BRs
Olivier

> Thank you for your help !!
> 
> Best regards
> ---
> Kuninori Morimoto
>