On Sun, Apr 23, 2023 at 09:30:11AM +0200, Takashi Iwai wrote:
On Sat, 22 Apr 2023 18:10:21 +0200, Oswald Buddenhagen wrote:
It's multi, not mono/stereo.
AFAICT, this doesn't do anything in the kernel.
... but those values are read by user-space.
Also, I think the subclass is meaningless for devices with just one stream, but whatever.
Again, the value is read by user-space.
i assumed that much. but these are another thing that appears to have exactly zero useful documentation.
So changing both have clear influence on the user-space program, and unless you have to change this for fixing a real bug (and there is no other way), this is too risky. IOW, too late to change, we have to accept those values.
there aren't any precedents for use of SNDRV_PCM_CLASS_MULTI and SNDRV_PCM_SUBCLASS_MULTI_MIX in the kernel tree. there don't appear to be relevant hits outside the kernel, either. it's conceivable that some code would check for the *_GENERIC enums, but i didn't find such code. so i'd postulate that these enums are effectively dead, and both the risk and the gain of this change are about zero. i suggest to initiate a formal deprecation procedure for the MULTI enum values, however that's supposed to look like.
regards