[BUG] Using Internal Speaker on Laptop with Tiger Lake-H, Audio is delayed when playing from silence

[Aikoyori]

Describe the bug When I try to play audio using my laptop (MSI GP76 Leopard 11UG with Intel i7-11800H), the audio is muted for around a second before playing, skipping the silent part. Only happens on Internal Speaker and does not happen on Headphones.

To Reproduce

1. Be on Tiger Lake
2. Play some music from silent

Reproduction Rate Every time I started playing any audio, even system beep (the water drop sound)

Expected behavior The sound plays normally without silence part

Impact It is in fact very annoying when I don't want to wear earbuds.

Environment

1. Branch name and commit hash of the 2 repositories: sof (firmware/topology) and linux (kernel driver).
   • Kernel: 6.6.9-arch1 (24a07cdb07452c7e631da9d3913ea5d3bddbc71f)
   • SOF: 2023.12-1 (from Arch Linux Repository)
2. Name of the topology file
   • Topology: i have no idea actually because I didn't do much configuration on sof side of things
3. Name of the platform(s) on which the bug is observed.
   • Platform: Arch Linux

Screenshots or console output

https://github.com/thesofproject/sof/assets/12034280/32e46aa7-f89b-4dff-87da-395e23835f19

https://github.com/thesofproject/sof/assets/12034280/6f5be9aa-b074-44c5-aee4-ad6a3d8618b4

[Aikoyori]

Also happens on my Ultramarine Linux 38 USB stick running alsa-sof-firmware 2.2.5-1.fc38 and Linux Kernel 6.3.5-200.fc38.x86_64

[plbossart]

Please follow the recommendations in https://thesofproject.github.io/latest/getting_started/intel_debug/suggestions.html#disable-sof-on-pci-hdaudio-devices-to-test-audio-playback and check if the problem happens with

options snd-intel-dspcfg dsp_driver=1

[Aikoyori]

I will try that out later today. Thanks.

[Aikoyori]

The problem still happens to a lesser degree (the delay is shorter but still exists).

[plbossart]

It's probably a slow ramp on the amplifier side, not an SOF bug proper but a codec driver configuration.

If you are using pulseaudio or pipewire the easiest 'fix' would be to disable the suspend-on-idle which would keep streaming zeroes into the audio device and prevent it from suspending. Not great for power but it would keep the missing initial audio.

Fixing the driver would require help from the codec vendor, or reverse-engineering the Windows setup https://thesofproject.github.io/latest/getting_started/intel_debug/suggestions.html#reverse-engineer-the-windows-audio-driver, which isn't straightforward and does not come with any guarantees of success.

[korewaChino]

Cannot reproduce on same chipset, Tried to reproduce on:

        *-multimedia
             description: Multimedia audio controller
             product: Tiger Lake-LP Smart Sound Technology Audio Controller
             vendor: Intel Corporation
             physical id: 1f.3
             bus info: pci@0000:00:1f.3
             logical name: card0
             logical name: /dev/snd/controlC0
             logical name: /dev/snd/hwC0D0
             logical name: /dev/snd/hwC0D2
             logical name: /dev/snd/pcmC0D0c
             logical name: /dev/snd/pcmC0D0p
             logical name: /dev/snd/pcmC0D31p
             logical name: /dev/snd/pcmC0D3p
             logical name: /dev/snd/pcmC0D4p
             logical name: /dev/snd/pcmC0D5p
             logical name: /dev/snd/pcmC0D6c
             logical name: /dev/snd/pcmC0D7c
             version: 20
             width: 64 bits
             clock: 33MHz
             capabilities: pm msi bus_master cap_list
             configuration: driver=sof-audio-pci-intel-tgl latency=32
             resources: iomemory:600-5ff iomemory:600-5ff irq:154 memory:6001138000-600113bfff memory:6001000000-60010fffff

Problematic machine:

*-multimedia
     description: Multimedia audio controller
     product: Tiger Lake-H HD Audio Controller
     vendor: Intel Corporation
     physical id: 1f.3
     bus info: pci@0000:00:1f.3
     logical name: card2
     logical name: /dev/snd/controlC2
     logical name: /dev/snd/hwC2D0
     logical name: /dev/snd/hwC2D2
     logical name: /dev/snd/pcmC2D0c
     logical name: /dev/snd/pcmC2D0p
     logical name: /dev/snd/pcmC2D31p
     logical name: /dev/snd/pcmC2D3p
     logical name: /dev/snd/pcmC2D4p
     logical name: /dev/snd/pcmC2D5p
     logical name: /dev/snd/pcmC2D6c
     logical name: /dev/snd/pcmC2D7c
     version: 11
     width: 64 bits
     clock: 33MHz
     capabilities: pm msi bus_master cap_list
     configuration: driver=sof-audio-pci-intel-tgl latency=32
     resources: iomemory:620-61f iomemory:620-61f irq:199 memory:6203148000-620314bfff memory:6203000000-62030fffff

The firmware version seems different, @Aikoyori's version is 11 while mine is 20. So probably a firmware issue. Any idea how you would update this?

[korewaChino]

Support package from chultrabook/linux-tools: debug-logs.tar.gz

[gwaewion]

I have the same issue with MSI GP66 Leopard 11UG with Intel i7-11800H. Also 'version: 11'.

[ujfalusi]

PA / PW should keep audio running for about 5 seconds between stop and start and streams 0 during that time.

Can you confirm that with newer kernel, firmware, topology and PA/PW you still have the issue or is it working better?

Usually an audio server improves the situation like this greatly.