Re: [PATCH v4 6/8] ASoC: compress: Add support for DAI multicodec

[Lars-Peter Clausen <lars@metafoo.de>]

On 07/03/2014 06:15 PM, Vinod Koul wrote:
> On Thu, Jul 03, 2014 at 01:18:46PM +0100, Mark Brown wrote:
>>> BUT Now thinking over this again, does this make sense do do with multiple
>>> codecs??
>>
>>> For folks haveing decoders in DSP inside SoC, the compressed device will be
>>> actually a FE. The BE will be PCM to which codec would be linked. The folks
>>> supporting decoders inside codec like WM wont ever need this.
>>
>>> Mark, do you agree to this?
>>
>>> So re-thinking again on why we need this??
>>
>> The use case tends to be for applications that have one device per
>> physical output.  This isn't normally mobile, it's normally high
>> performance audio applications where people are doing things to try to
>> electrically isolate the analogue outputs or dealing with high power so
>> need to keep speaker outputs physically separate.  I can imagine set top
>> box type applications (which do use offloaded media decode) doing this.
>
> Not sure if I follow you, I have no idea of how these systems work.
>
> But if sound card has compressed device it would need to represent using DPCM as
> we don't know the decoder PCM output. The multiple codecs would then connect to
> the PCM BE.
> So on Compressed FE, we wont have multiple codecs.
>
> For codecs, the compressed audio will go to one codec, multiple ones wont make
> sense.
>
> So how can we have a situation where we have compressed device linked to
> multiple codecs?
>

If the assumption is that a DAI link that has a compressed DAI on the CPU side 
is never connected to a 'real' CODEC (or multiple CODECS) on the CODEC side. I 
think we can just leave soc-compress.c alone and just add a sanity check to 
make sure that num_codecs is always 1.

- Lars