[BUG] microphone detected but not working on 2022 Razer Blade 15

[ewfuentes]

Describe the bug
I've installed Ubuntu 22.04 on a 2022 Razer Blade 15. Audio playback appears to be working (apart from L/R audio channels being flipped), but the built in microphone, while detected, doesn't seem to be recording any sound. This was tested by going to the ubuntu sound settings, making some noise and not seeing any activity from the meter.

I'm unfamiliar with linux sound, but I'm willing to do some digging in. Thank you for your time!

To Reproduce
Clean install of Ubuntu 22.04. I have updated the kernel to 5.17.3 to get wifi support.

Reproduction Rate
100%

Expected behavior
I expect the microphone to record when enabled.

Impact
I can't use the built in microphone for video conferencing. I haven't tried wired/bluetooth or USB microphones as a workaround just yet.

Environment

1. Branch name and commit hash of the 2 repositories: sof (firmware/topology) and linux (kernel driver).
   • Kernel: f6a218ea3564d2c2e53199577929e02e173a499c
   • SOF: I'm not quite sure. I've installed firmware-sof-signed 2.0-1ubuntu3
2. Name of the topology file
   • Topology: I'm not sure how to determine this... I've included a link to the alsa-info artifact, but nothing sticks out to me as having this information
3. Name of the platform(s) on which the bug is observed.
   • Platform: x86_64

Screenshots or console output

http://alsa-project.org/db/?f=625e3480bfdd42663c730e65229c8619950e749c

[plbossart]

There's no smoking gun @ewfuentes, the only part where we have a known issue is the case of 2 DMICs connected to PDM1, instead of the usual PDM0.

Can you please add this to /etc/modprobe.d/alsa-base.conf

options snd_sof_intel_hda_common dmic_num=4

reboot and run this command

arecord -l

check that the DMIC device is card 0, device 6, then

arecord -Dhw:0,6 -c4 -r48000 -sS32_LE -d 10 test.wav

Try to record your voice during this 10s test, and then look at the test.wav file with audacity.

If you see any signs of life on 2 channels, we have a mic configuration issue. If you see nothing, then something else is at play.

Hope this helps.

[ewfuentes]

After adding the option to alsa-base.conf, arecord -l shows the microphone as being on card 1 (instead of 0) device 6.

**** List of CAPTURE Hardware Devices ****
card 1: sofhdadsp [sof-hda-dsp], device 0: HDA Analog (*) []
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 1: sofhdadsp [sof-hda-dsp], device 1: HDA Digital (*) []
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 1: sofhdadsp [sof-hda-dsp], device 6: DMIC (*) []
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 1: sofhdadsp [sof-hda-dsp], device 7: DMIC16kHz (*) []
  Subdevices: 1/1
  Subdevice #0: subdevice #0

I used the following command to record myself (note the device and -f instead of -s):

arecord -Dhw:1,6 -c4 -r48000 -fS32_LE -d 10 test.wav

Looking at the wav file in audacity shows some promise! There are 4 channels and channels 3 and 4 contain me speaking. Channels 1 and 2 have some audio artifact , but are otherwise silent.

Screenshot of the wav file:

Screenshot of the audio artifact:

For what it's worth, the audio meter in the sound settings and google meet don't appear to recognize any audio.

[plbossart]

ok, so please remove the 'options' from /etc/modprobe.d/alsa-base.conf and follow the guidance in this link: thesofproject/linux#2460 (comment)

"
The solution for Lenovo devices is to try to use the topology provided in https://github.com/thesofproject/linux/files/5981682/sof-hda-generic-2ch-pdm1.zip

Unzip and copy sof-hda-generic-2ch-pdm1.tplg over /lib/firmware/intel/sof-tplg/sof-hda-generic-2ch.tplg (KEEP THE INITIAL VERSION AS A BACKUP).
"

[ewfuentes]

That did the trick! Thank you so much for your help! Is there anything else that needs to happen? Is there some way I can help fix this for others?

[plbossart]

excellent @ewfuentes

Can you do me a favor and retest with this file
sof-hda-generic-2ch-pdm1.tplg.gz

it should be the same functionality as before, just trying to make sure the updated file is correct.

[ewfuentes]

Should I copy it over as sof-hda-generic-2ch.tplg, or should I paste it with it's existing name?

[plbossart]

copy it over as sof-hda-generic-2ch.tplg

[ewfuentes]

hmmm... That one didn't seem to work. Here's what I've got for hashes:

erick@neptune:~/Downloads$ md5sum *.tplg 
d3d7af694582142e087452e845817ed2  sof-hda-generic-2ch-pdm1-orig.tplg <--- The one from the lenovo thread
df3fab3ec07c925e391deae8eeb4363f  sof-hda-generic-2ch-pdm1.tplg <--- The last one you sent me
3a0f76a6e276b9d0cb10488566fdcf4b  sof-hda-generic-2ch.tplg <--- The one installed by default
erick@neptune:~/Downloads$ sudo md5sum /lib/firmware/intel/sof-tplg/sof-hda-generic-2ch.tplg 
[sudo] password for erick: 
df3fab3ec07c925e391deae8eeb4363f  /lib/firmware/intel/sof-tplg/sof-hda-generic-2ch.tplg

let me try reverting to the previous one.

[ewfuentes]

Confirming that the original topology you pointed me to works (d3d7af69). Whereas the new one does not (df3fab3e). This was verified by trying to record in Audacity.

[plbossart]

Can you try with this update
sof-hda-generic-2ch.tplg.gz

I only hard-coded a PDM0 to PDM1 change, if that doesn't work we'll have to ask @singalsu for help.

diff --git a/tools/topology/topology1/platform/intel/intel-generic-dmic.m4 b/tools/topology/topology1/platform/intel/intel-generic-dmic.m4
index c9808c3d9..5dfef3a5b 100644
--- a/tools/topology/topology1/platform/intel/intel-generic-dmic.m4
+++ b/tools/topology/topology1/platform/intel/intel-generic-dmic.m4
@@ -157,7 +157,7 @@ ifelse(DMIC_DAI_CHANNELS, 4,
 `DAI_CONFIG(DMIC, 0, DMIC_DAI_LINK_48k_ID, DMIC_DAI_LINK_48k_NAME,
            DMIC_CONFIG(1, 2400000, 4800000, 40, 60, 48000,
                 DMIC_WORD_LENGTH(s32le), 200, DMIC, 0,
-                PDM_CONFIG(DMIC, 0, STEREO_PDM0)))')
+                PDM_CONFIG(DMIC, 0, STEREO_PDM1)))')
 
 ifdef(NO16KDMIC, `',
 `ifelse(DMIC16K_DAI_CHANNELS, 4,
@@ -168,4 +168,4 @@ ifdef(NO16KDMIC, `',
 `DAI_CONFIG(DMIC, 1, DMIC_DAI_LINK_16k_ID, DMIC_DAI_LINK_16k_NAME,
            DMIC_CONFIG(1, 2400000, 4800000, 40, 60, 16000,
                 DMIC_WORD_LENGTH(s32le), 400, DMIC, 1,
-                PDM_CONFIG(DMIC, 1, STEREO_PDM0)))')')
+                PDM_CONFIG(DMIC, 1, STEREO_PDM1)))')')

There are additional changes in the config that I don't quite get, the old working file has this
DMIC_CONFIG(1, 500000, 4800000, 40, 60, 48000,
and the newer files:
DMIC_CONFIG(1, 2400000, 4800000, 40, 60, 48000,

[ewfuentes]

The one you just sent me (dd17ba1a) does work.

erick@neptune:~/Downloads$ sudo md5sum /lib/firmware/intel/sof-tplg/sof-hda-generic-2ch.tplg 
dd17ba1ab213a6b1827334ca83a382a4  /lib/firmware/intel/sof-tplg/sof-hda-generic-2ch.tplg
erick@neptune:~/Downloads$ md5sum *.tplg*
3a0f76a6e276b9d0cb10488566fdcf4b  sof-hda-generic-2ch.tplg.bak <--- originally installed
d3d7af694582142e087452e845817ed2  sof-hda-generic-2ch-pdm1-orig.tplg <--- from lenovo thread
df3fab3ec07c925e391deae8eeb4363f  sof-hda-generic-2ch-pdm1.tplg <--- First one you sent me that didn't work
dd17ba1ab213a6b1827334ca83a382a4  sof-hda-generic-2ch.tplg <--- second one you sent me that does work.

[plbossart]

ok, that's good @ewfuentes. I had a mistake in my M4 file, fixed now.

So the short story is that you have a working solution. If you want a long-term solution, we will have this topology file included in official releases (2.3+) and you will be able to use a kernel parameter to select the topology (no file override required). Unfortunately, we have a number of bugs in the kernel override so you will need additional patches from thesofproject/linux#3740.