[alsa-devel] [PATCH RFC v2 2/4] ASoC: Add multiple CPU DAI support for PCM ops

Pierre-Louis Bossart pierre-louis.bossart at linux.intel.com
Thu Jan 16 03:25:02 CET 2020


>>> +	/* Do we need to support Multi cpu for FE? */
>>> +	for_each_rtd_cpu_dai(fe, i, fe_cpu_dai) {
>>> +		if (fe_cpu_dai->active) {
>>> +			err = soc_pcm_apply_symmetry(fe_substream,
>> fe_cpu_dai);
>>> +			if (err < 0)
>>> +				return err;
>>> +		}
>>
>> that part seems inconsistent with Patch 4 where you add warnings/error
>> everywhere there's a FE with num_cpus>1
> 
> Actually, I have a question here. Do we need to support Multi cpu for FE?
> If yes, what is the use case?

I can't think of any case where we'd need a multi-cpu FE, and if you 
look at the existing code it doesn't look possible:

we have code in soc-dapm.c that checks that num_cpus==num_codecs and 
flags an error if you had 2 cpu_dais and 1 codec_dai.

And if you continue with soc-topology, we have hard-coded num_cpus = 1; 
and num_codecs = 1; for topology-created FE dailinks.

In other words, we'd a major topology change to support a multi-cpu FE, 
quite unlikely for now.

The multi-cpu representation is IMO only needed for specific hardware 
setups, just like the multi-codec part is tied to TDM usages.


More information about the Alsa-devel mailing list