[juimonen]

Currently we are checking if matrix row contains any 1's (popcount). In
mux case this doesn't tell us if channels are mixed. For example if we
have 2 input matrix in which first row's position 2 and 4 have value 1,
it means that both streams channel 1 is copied to output channels 2 and
4, so not mixed. So change the check to be channel specific.

Signed-off-by: Jaska Uimonen jaska.uimonen@intel.com

[juimonen]

@bkokoszx @libinyang FYI

[libinyang]

@juimonen when I apply your patch, It shows the below warning. I know we can use --no-verify to skip this warning. But it should be better to fix this warning, right?

WARNING: Too many leading tabs - consider code refactoring
#31: FILE: src/audio/mux/mux.c:91:
+                                               if (!channel_set) {

WARNING: Too many leading tabs - consider code refactoring
#33: FILE: src/audio/mux/mux.c:93:
+                                               } else {

[bkokoszx]

@juimonen
Thanks for this PR. I have one concern. Do we also check whether we perform mixing inside each individual input stream? I mean that we have e.g.:
cfg->streams[0].mask[0] = 0x1
cfg->streams[0].mask[1] = 0x1
It means that 0 and 1 channel from 0 stream goes to first channel at output, so there are also mixed. I think we do not check this case ATM.

[juimonen]

@bkokoszx yes you are right, let me try to make separate check functions for individual matrix and "inter-matrix", they are little bit nasty to combine and checkpatch goes also crazy for intendation level.

[juimonen]

@libinyang @bkokoszx new version, can you help to test and review?

[bkokoszx]

Now, looks good to me. Thanks

[lgirdwood]

@juimonen can you check the CI

[juimonen]

SOFCI TEST

[juimonen]

updated: removed some rimage related bogus and added general comment to the check function