Hi Jerome
Thank you for your feedback
DPCM [CPU/xxxx]-[xxxx/Codec] ^^^^ (A)
(snip)
DPCM [CPU/xxxx]-[xxxx/Codec] ^^^^^^^^^ ^^^^^^^^^^ (B)
(snip)
if ((dai_link->no_pcm) &&
^ Actually my comment applies to all links, DPCM backend or not
(snip)
A codec that does not support playback and does not support capture does not support much, does it ? ;)
Unfortunately, some codec which is used on DPCM BE doesn't have .channels_min = 1 settings which is used on snd_soc_dai_stream_valid(). At least Intel is using it. For both Playback/Capture.
But it *was* no problem, because (A) part was not checked before.
Because of this background, [01/15] patch is using dummy Codec to avoid (B) check.
[15/15] patch will indicate WARNING for such codec driver instead of just avoiding. At least, below are the drivers. It is mentioned in git-log.
sound/soc/codecs/hda.c sound/pci/hda/patch_hdmi.c
dai_link->no_pcm only is OK I think, because it needs to keep compatibility and try to indicates warning for (A) and/or (B) part. If such things happen on FE side, it is just error, not warning.
(!cpu_play_t && !codec_capture_t)) {Then at first glance, maybe ... CPU and codec seem to exclude each other but that will only work as long as DCPM is limited to a single CPU per link.
Hmm ?? I'm confusing Did you copy-and-paste miss ?? I have never indicate this pair
+ I have indicated before - I have not
+ ( cpu_capture_t && !codec_capture_t) // in [15/15] patch + (!codec_play_t && !codec_capture_t) // in previous mail - (!cpu_play_t && !codec_capture_t)
and I'm sorry but I can't understand what you want to tell me. Do you mean Multi-CPU case ? If you can indicate Image or code, it can more help me.
Thank you for your help !!
Best regards --- Renesas Electronics Ph.D. Kuninori Morimoto