[BUG] [MTL] Playback clips captured from mock RTC AEC device is double speeded

[yongzhi1]

Describe the bug

Playback the file captured from "device 27: DMIC0 RTC AEC" with CONFIG_GOOGLE_RTC_AUDIO_PROCESSING_MOCK=y, and hear the speed is doubled.

To Reproduce

Start spk playback:

aplay -Dhw:0,1 -c2 -r48000 -fS16_LE /home/LR.wav -vv

Start spk feedback capture:

arecord -Dhw:0,29 /tmp/tmp.wav -c2 -r48000 -fs32_le -vv

Start AEC capture:

arecord -Dhw:0,27 /tmp/mock_aec.wav -c4 -r48000 -fs32_le -vv -d20

Reproduction Rate
100%

Environment

1. Branch name and commit hash of the 2 repositories: sof (firmware/topology) and linux (kernel driver).

   • Kernel: {https://chromium.googlesource.com/chromiumos/third_party/kernel/+/refs/heads/chromeos-6.1}
   • SOF: {MTL-005.0.3 release} https://github.com/thesofproject/sof/releases/tag/releases%2Fmtl%2Fv5.0.3
   • rimage: RanderWang/rimage@bbdbf09

2. Name of the topology file

   • Topology: {sof-mtl-max98357a-rt5682-ssp2-ssp0.tplg} plus
     topology2: fixup google aec topology #7970

3. Name of the platform(s) on which the bug is observed.

   • Platform: {MTL/Rex}

Please also include the relevant sections from the firmware log and kernel log in the report (and attach the full logs for complete reference).

mtrace.txt
dmesg-mock-aec.txt
mock_aec.zip

[lgirdwood]

@yongzhi1 can you confirm if its SPK feedback or AEC that is double speed or both ?
@mwasko @abonislawski fyi.

[yongzhi1]

Hi, @lgirdwood , to confirm, both spk playback and EchoRef ( spk loopback for capture) work normally w/o speed or glitch issue, thanks!

[abonislawski]

Please verify on newest chromeos, I cannot reproduce on main and 005

[RDharageswari]

hi Adrian,
i can repro the issue still on mtl-005

[serhiy-katsyuba-intel]

Looks like the problem is caused by wrong host-copier.27.capture output format configuration. It is configured to receive 2-channel audio on source and output 4-channel audio to host. See dmesg log:

[  666.200835] sof-audio-pci-intel-mtl 0000:00:1f.3: Init input audio formats for host-copier.27.capture
[  666.200836] sof-audio-pci-intel-mtl 0000:00:1f.3: Pin index #0: 48000Hz, 32bit (ch_map 0xffffff10 ch_cfg 1 interleaving_style 0 fmt_cfg 0x2002) buffer size 384
[  666.200837] sof-audio-pci-intel-mtl 0000:00:1f.3: Output audio format for host-copier.27.capture
[  666.200838] sof-audio-pci-intel-mtl 0000:00:1f.3: Pin index #0: 48000Hz, 32bit (ch_map 0xffff3210 ch_cfg 4 interleaving_style 0 fmt_cfg 0x2004) buffer size 768
[  666.200839] sof-audio-pci-intel-mtl 0000:00:1f.3: copier host-copier.27.capture, IPC size is 84
[  666.200839] sof-audio-pci-intel-mtl 0000:00:1f.3: COPIER (UUID: 9BA00C83-CA12-4A83-943C-1FA2E82F9DDA): No CPC match in the firmware file's manifest (ibs/obs: 384/768)
[  666.217456] sof-audio-pci-intel-mtl 0000:00:1f.3: host-copier.27.capture: ibs / obs / cpc: 384 / 768 / 0

Current version of copier does not support remapping of channels. So, basically, host-copier.27.capture just copy its input data to output DMA buffer and so host receives 2 times less data when expected.

Is it a mistake that host-copier.27.capture is configured with 4-channel output instead of 2? Or is this intentional and for some reason 4 channels are required?

cc: @ranj063.

[lgirdwood]

@RDharageswari I'm assuming the 4 channel output is not needed by userspace here ?

[ranj063]

@yongzhi1 @RDharageswari Do you still see this issue with #8101

[yongzhi1]

Hi, @ranj063 with #8101 and the mock module, the speed issue is still there when 4xch is enabled. The speed issue is not reproducible with 2xch DMIC (i.e. sof-mtl-max98357a-rt5682-ssp2-ssp0-2ch-pdm1). Thanks @serhiy-katsyuba-intel for the hint!!

[ranj063]

@serhiy-katsyuba-intel the current aec pipeline in topology is missing support for 4ch capture. I am working with @yongzhi1 to fix this.

[yongzhi1]

With #8101, #8172 and thesofproject/linux#4550, the issue can be closed, thanks @ranj063 for the prompt support!!