I downloaded this PR and ran it on Raven.
I noticed that the downmix is clipping, which is not the fault of this PR.
I also had a problem when selecting output channelMask of 5.1.4.
Pressing OPEN and START resulted in a crash. I think this is because the sine player can only handle up to 8 channels. So an array is getting over-indexed somewhere.

For TEST INPUT, if I select ChannelCount=3 then I get 3 independent mic inputs.
If I select any channelMask with >=3 bits set, e.g. "2.1", "Tri", then I get an error. Maybe this is WAI.

For TEST INPUT, if I select ChannelCount=3 then I get 3 independent mic inputs. If I select any channelMask with >=3 bits set, e.g. "2.1", "Tri", then I get an error. Maybe this is WAI.

For input case, "2.1" and "Tri" are not valid selection. In that case, getting error is WAI. Selecting channel count as 3 will be translated to index mask, which is a different case and will work.

For input case, "2.1" and "Tri" are not valid selection.

Is there a case when they would be valid? I am guessing no.

system/audio.h distinguishes between AUDIO_CHANNEL_IN_* and AUDIO_CHANNEL_OUT_*.
I don't think there is much use case for the input channel masks.
Maybe this can be handled with documentation.

I downloaded this PR and ran it on Raven. I noticed that the downmix is clipping, which is not the fault of this PR. I also had a problem when selecting output channelMask of 5.1.4. Pressing OPEN and START resulted in a crash. I think this is because the sine player can only handle up to 8 channels. So an array is getting over-indexed somewhere.

Yes, our previous max channel count is 8. I have uploaded a new patch to address this.

For input case, "2.1" and "Tri" are not valid selection.

Is there a case when they would be valid? I am guessing no.

system/audio.h distinguishes between AUDIO_CHANNEL_IN_* and AUDIO_CHANNEL_OUT_*. I don't think there is much use case for the input channel masks. Maybe this can be handled with documentation.

Yes, audio.h distinguishes between input and output channel. But when AIDL is introduced, we decided not to distinguish input and output channel mask AAudioStreamParameters::validateChannelMask for the valid input channel mask. For instance, 2.0.2 is a valid one. We can add that in the aaudio documentation.