[JSL][bug]"Dummy output" display in sound setting when power on with DP connected.

[sinahuang]

Describe the bug
"Dummy output" display in sound setting when power on with DP connected. check "amixer contents" the DP status is on. Aplay via DP is ok.

To Reproduce
1.Boot up system with DP connected.
2.Open sound setting to check after system boot up successfully.

Reproduce rate
5/5

Expected result
System should display output source normally in sound setting.

Actual result
"Dummy output" display in sound setting when power on with DP connected.

Test recipe
Kernel:https://github.com/thesofproject/linux/commits/topic/sof-dev commit: 1b2f260 +PR #2208
FW: https://github.com/thesofproject/sof/commits/master commit: 3d8a0ac
Tplg: sof-hda-generic-4ch.tplg
Platform:JSL-RVP with ALC700 in HDA mode

dmesg11.log
amixer19.txt
sof_trace11.log

[bardliao]

Something doesn't look right below. @sinahuang Could you also attach dmesg log while aplay via DP?

[    4.723734] sof-audio-pci 0000:00:1f.3: sink HDA1.IN control none source codec0_in
[    4.723740] sof-audio-pci 0000:00:1f.3: warning: neither Linked source component codec0_in nor sink component HDA1.IN is of buffer type, ignoring link
[    4.723745] sof-audio-pci 0000:00:1f.3: sink HDA3.IN control none source codec1_in
[    4.723749] sof-audio-pci 0000:00:1f.3: warning: neither Linked source component codec1_in nor sink component HDA3.IN is of buffer type, ignoring link
[    4.723754] sof-audio-pci 0000:00:1f.3: sink codec0_out control none source HDA0.OUT
[    4.723761] sof-audio-pci 0000:00:1f.3: sink codec1_out control none source HDA2.OUT
[    4.723768] sof-audio-pci 0000:00:1f.3: sink HDA3.IN control none source codec2_in
[    4.723771] sof-audio-pci 0000:00:1f.3: warning: neither Linked source component codec2_in nor sink component HDA3.IN is of buffer type, ignoring link
...
[    8.550823] snd_hda_codec_hdmi ehdaudio0D2: HDMI: detected monitor DELL U2414H at connection type DisplayPort
[    8.550829] snd_hda_codec_hdmi ehdaudio0D2: HDMI: available speakers: FL/FR
[    8.550836] snd_hda_codec_hdmi ehdaudio0D2: HDMI: supports coding type LPCM: channels = 2, rates = 32000 44100 48000, bits = 16 20 24
[    8.550841] snd_hda_codec_hdmi ehdaudio0D2: HDMI: detected monitor DELL U2414H at connection type HDMI
[    8.550844] snd_hda_codec_hdmi ehdaudio0D2: HDMI: available speakers: FL/FR
[    8.550847] snd_hda_codec_hdmi ehdaudio0D2: HDMI: supports coding type LPCM: channels = 2, rates = 32000 44100 48000, bits = 16 20 24
[    8.550867] snd_hda_codec_hdmi ehdaudio0D2: HDMI: hinfo 00000000b53bab05 not registered
...
[    8.558870] snd_hda_codec_hdmi ehdaudio0D2: hdmi_pin_hbr_setup: NID=0x6, pinctl=0x40
[    8.558993] snd_hda_codec_hdmi ehdaudio0D2: hda_codec_setup_stream: NID=0x5, stream=0x2, channel=0, format=0x31
[    8.559630] snd_hda_codec_hdmi ehdaudio0D2: ASoC: error at snd_soc_dai_startup on intel-hdmi-hifi1: -16
[    8.559638] snd_hda_codec_hdmi ehdaudio0D2: ASoC: can't open DAI intel-hdmi-hifi1: -16
[    8.559644]  iDisp1: ASoC: BE open failed -16
[    8.559647]  HDMI1: ASoC: failed to start some BEs -16

[sinahuang]

Error still can be seen with PR #2223.

[sinahuang]

Attached log while aplay via DP:
dmesg_aplay.log

[kv2019i]

Hmm, that is interesting .. it seems i915 is disabling audio codec while SOF audio probe is still ongoing, resulting in this:
May 28 21:00:55 sh-Jasper-Lake-HDA kernel: [ 5.029048] WARNING: CPU: 2 PID: 266 at sound/pci/hda/hda_codec.c:2537 snd_hda_spdif_ctls_unassign+0x45/0x50 [snd_hda_codec]
...
May 28 21:00:55 sh-Jasper-Lake-HDA kernel: [ 5.029089] update_eld+0x6e9/0x7c0 [snd_hda_codec_hdmi]
May 28 21:00:55 sh-Jasper-Lake-HDA kernel: [ 5.029090] ? i915_audio_component_get_eld+0x139/0x140 [i915]
May 28 21:00:55 sh-Jasper-Lake-HDA kernel: [ 5.029091] hdmi_present_sense+0x16a/0x1e0 [snd_hda_codec_hdmi]

This shouldn't lead to critical failures though. Could you share "sudo alsa-info --stdout" output and pulseaudio verbose log that covers start-up and the use-case sequence?

sudo mv /usr/bin/pulseaudio /usr/bin/pulseaudio-off
killall -v pulseaudio
pulseaudio-off -vvv &> pulse-log.txt
# exit with ctrl-c)
sudo mv /usr/bin/pulseaudio-off /usr/bin/pulseaudio

[sinahuang]

Attached the alsa-info output and pluseaudio logs:
pulse-log1.txt
alsa-info1.txt

[kv2019i]

Thanks @sinahuang ! It seems I'm still missing a case in PR2223 and pulseaudio init fails:
I: [pulseaudio-off] alsa-util.c: Error opening PCM device hw:sofhdadsp,3: Device or resource busy
D: [pulseaudio-off] alsa-mixer.c: Profile set 0x560925c144e0, auto_profiles=no, probed=yes, n_mappings=0, n_profiles=0, n_decibel_fixes=0
E: [pulseaudio-off] module-alsa-card.c: Failed to find a working profile.

... I'll work on a fix.

[kv2019i]

I have a hypothesis this is caused by intel_not_share_assigned_cvt() function not working as expected. Hint provided in the log:

May 28 21:00:58 sh-Jasper-Lake-HDA kernel: [ 8.724401] snd_hda_codec_hdmi ehdaudio0D2: choose cvt 1 for pin nid 4

[kv2019i]

Same rootcause as #2216 (comment) , will ETA next week.