[RFC PATCH v2 4/6] ASoC: fsl_asrc_m2m: Add memory to memory function

[Pierre-Louis Bossart]

On 8/20/24 09:42, Jaroslav Kysela wrote:
> On 20. 08. 24 9:37, Shengjiu Wang wrote:
>> On Tue, Aug 20, 2024 at 2:59 PM Pierre-Louis Bossart
>> <pierre-louis.bossart at linux.intel.com> wrote:
>>>
>>>
>>>
>>> On 8/20/24 04:53, Shengjiu Wang wrote:
>>>> On Mon, Aug 19, 2024 at 3:42 PM Pierre-Louis Bossart
>>>> <pierre-louis.bossart at linux.intel.com> wrote:
>>>>>
>>>>>
>>>>>
>>>>> On 8/16/24 12:42, Shengjiu Wang wrote:
>>>>>> Implement the ASRC memory to memory function using
>>>>>> the compress framework, user can use this function with
>>>>>> compress ioctl interface.
>>>>>>
>>>>>> Define below private metadata key value for output
>>>>>> format, output rate and ratio modifier configuration.
>>>>>> ASRC_OUTPUT_FORMAT 0x80000001
>>>>>> ASRC_OUTPUT_RATE   0x80000002
>>>>>> ASRC_RATIO_MOD     0x80000003
>>>>>
>>>>> Can the output format/rate change at run-time?
>>>>
>>>> Seldom I think.
>>>>
>>>>>
>>>>> If no, then these parameters should be moved somewhere else - e.g.
>>>>> hw_params or something.
>>>>
>>>> This means I will do some changes in compress_params.h, add
>>>> output format and output rate definition, follow Jaroslav's example
>>>> right?
>>>
>>> yes, having parameters for the PCM case would make sense.
>>>
>>>>> I am still not very clear on the expanding the SET_METADATA ioctl to
>>>>> deal with the ratio changes. This isn't linked to the control layer as
>>>>> suggested before, and there's no precedent of calling it multiple
>>>>> times
>>>>> during streaming.
>>>>
>>>> Which control layer? if you means the snd_kcontrol_new?  it is bound
>>>> with sound card,  but in my case,  I need to the control bind with
>>>> the snd_compr_stream to support multi streams/instances.
>>>
>>> I can only quote Jaroslav's previous answer:
>>>
>>> "
>>> This argument is not valid. The controls are bound to the card, but the
>>> element identifiers have already iface (interface), device and subdevice
>>> numbers. We are using controls for PCM devices for example. The binding
>>> is straight.
>>>
>>> Just add SNDRV_CTL_ELEM_IFACE_COMPRESS define and specify the compress
>>> device number in the 'struct snd_ctl_elem_id'.
>>> "
>>
>> I don't think it is doable,  or at least I don't know how to do it.
>>
>> My case is just one card/one device/one subdevice.  can't use it to
>> distinguish multi streams.
> 
> I already wrote that the compress core code should be extended to
> support subdevices like other ALSA APIs. I'll work on it. For now, just
> add support for one converter.

I am not sure I get the benefits of subdevices in this context.

Can we not use different devices already, one per hardware ASRC instance?

Put differently, what would be the difference between a card with N
compressed devices or a card with 1 compressed device and N subdevices?