Repository navigation
[BUG] sof-hda-generic-1ch.tplg not found #9748
Description
Describe the bug
No Soundcards are initialised.
To Reproduce
Boot the System (ASUS NUC14LNS)
Reproduction Rate
100%
Expected behavior
Topology is loaded, and soundcard is initialised
Impact
No audio
Environment
- Branch name and commit hash of the 2 repositories: sof (firmware/topology) and linux (kernel driver).
- Kernel: Linux version 6.12.0-next-20241127-test-kernel (dd@dd-ASUS-EXPERTBOOK) (gcc (Ubuntu 14.2.0-4ubuntu2) 14.2.0, GNU ld (GNU Binutils for Ubuntu) 2.43.1) Cavs common #4 SMP PREEMPT_DYNAMIC Wed Nov 27 17:53:05 CST 2024
- SOF: v2.11.1
- Name of the topology file
- Topology: sof-hda-generic-1ch.tplg
- Name of the platform(s) on which the bug is observed.
- Platform: Intel(R) Core(TM) Ultra 9 288V 3.30GHz
- ASUS NUC14LNS
Screenshots or console output
Initial log with missing file:
NUC14LNS_dmesg.txt
I also attempted to generate sof-hda-generic-1ch.tplg by modifying tools/topology/topology2/production/tplg-targets-hda-generic.cmake, to add the configuration where NUM_DMICS=1:
///"sof-hda-generic;sof-hda-generic-1ch;HDA_CONFIG=mix,NUM_DMICS=1,
DMIC0_ENHANCED_CAPTURE=true,EFX_DMIC0_TDFB_PARAMS=line2_generic_pm10deg,
EFX_DMIC0_DRC_PARAMS=dmic_default"///
But this didn't work due, producing these error messages:
[ 7.026064] DMIC Raw: substream DMIC Raw has no playback, no capture
[ 7.026071] skl_hda_dsp_generic skl_hda_dsp_generic: ASoC: can't create pcm DMIC Raw :-22
[ 7.026722] skl_hda_dsp_generic skl_hda_dsp_generic: probe with driver skl_hda_dsp_generic failed with error -22
Full log attached here:
1219_dmesg.txt
Activity
kv2019i commentedon Dec 19, 2024
This is a curious case, I don't believe we've seen a single dmic configuration before. @singalsu @bardliao @ujfalusi do you recall?
@stefdb Can quick workaround to try is to force the dmic count to 2 with:
/etc/modprobe.d/sof.conf (or some file under modprobe.d):
This should instruct the drive to load sof-hda-generic-2ch.tplg which is part of standard distribution. If this works, you'll probably have audio only on the left channel of the recorded stream, but at least allows audio device to probe successfully.
bardliao commentedon Dec 20, 2024
No, it is the first time I am aware of that.
singalsu commentedon Dec 30, 2024
First time for me too. Please use as @kv2019i suggested the 2ch topology as replacement. Due to PDM bus characteristic, it will give a dual mono capture stream as e.g. HDA generic headset (mono) mic also appears to system.
I will try to replicate the issue with my devices.
@stefdb Can you please also dump the NHLT content of your PC and attach here (
sudo cat /sys/firmware/acpi/tables/NHLT >NHLT.bin).singalsu commentedon Jan 2, 2025
Confirming that I'm able to replicate the issue by forcing
options snd_sof_intel_hda_generic dmic_num=1plus andoptions snd_sof_intel_hda_common sof_use_tplg_nhlt=1and similarly creating a topology sof-hda-generic-cavs25-1ch.tplg renamed to sof-hda-generic-1ch.tplg. I get the same error as reported.singalsu commentedon Jan 3, 2025
Update - I'm testing a fix in #9763.
stefdb commentedon Jan 7, 2025
NHLT.zip
Hi,
I've attached NHLT.bin (inside NHLT.zip) from the system, as requested.
Thanks,
Stefan
singalsu commentedon Jan 7, 2025
Thanks a lot for sharing the NHLT @stefdb !
It looks like there's an issue in DMIC HW setting. The NHLT headers show 1 channel but some actual DMIC decimators settings activate parts of HW as stereo. If there's a problem we should contact Asus.
Can you first remove
options snd_sof_intel_hda_generic dmic_num=2from /etc/modprobe.d/sof.conf (or what you used) and trysof-hda-generic-1ch.tplgfrom the attached tar bundle. Check if the SOF boots and if capture produces a sane recording (no noise, no mickey mouse or stretched deep voice effect). There's a big risk that the recording might not be good quality if it can run.I ran on my test device the capture (had forced dmic_num=1 on a 4ch device) like this:
time arecord -Dhw:0,6 -f S32_LE -r 48000 -c 1 -d 10 rec.wavThe capture should take about 10s.
There's a way to override the BIOS NHLT and use NHLT from topology (a method used in Chromebooks). Add these options to e.g. /etc/modprobe.d/sof.conf and then try with
sof-hda-generic-ace1-1ch.tplgfrom the bundle. If the first way failed, this should always work.The topologies were generated with #9763.
(removed the files bundle because of issue in ace1/ace3 version)
singalsu commentedon Jan 7, 2025
@stefdb Please wait testing attached sof-hda-generic-ace1-1ch.tplg. I don't have suitable MTL/LNL hardware to test this so I ran it here with the cAVS version that worked. There may be a bug in alsa-utils topology NHLT generator for ACE. I will do more checks and try to fix it.
singalsu commentedon Jan 8, 2025
Here's new topologies bundle (fix in alsa-project/alsa-utils#286). Please try sof-hda-generic-ace1-1ch.tplg from here:
dmic_1ch_tplgs_20250108.tar.gz
lgirdwood commentedon Jan 14, 2025
@stefdb any update ? did this work for you ?
stefdb commentedon Jan 16, 2025
Hi,
We've tested these topologies and the dmic seems to work now, and the soundcard probes.
However, the speaker is not currently working, and I've been trying to work out why.
Previously we tested with SOF disabled, and the speaker worked ok.
The strange thing is that it seems like the play command isn't being propagated to the kernel, so it seems something in userspace is not happy, but I'm still investigating.
I've attached 2 dmesg logs of using the new topologies.
Thanks,
Stefan
20240113_generic-ace1-1ch.txt
20250113_generic-1ch.txt
singalsu commentedon Jan 17, 2025
@stefdb Great that the microphone works now! Please create another issue for the speaker and attach the above logs there. Especially if it needs help from Cirrus (some fw upload fails seen in dmesg), it's more clear to handle it there.
singalsu commentedon Jan 17, 2025
We can close this issue after these are merged and verified:
bardliao commentedon Jan 17, 2025
@stefdb Yes, please create another issue for the speaker issue for better tracking. Please also attach the output of
alsa-info.And what is your playback command? Did you try aplay?
lgirdwood commentedon Feb 10, 2025
@singalsu can we close now ?
singalsu commentedon Feb 11, 2025
Yep, everything related to DMIC is now completed. Please continue in other issue with speaker problem.