Skip to content

DMIC test fail when audio could not play #800

Closed
Closed
DMIC test fail when audio could not play#800
Assignees
keyonjiemmaka1slawblauciakmichalgrodzicki
Labels
GLKApplies to Gemini Lake platformP1Blocker bugs or important featuresbugSomething isn't working as expectedduplicateThis issue or pull request already exists

Description

@naveen-manohar
naveen-manohar
opened on Jan 8, 2019

Codebase: ChromeOS-firmware-R72-11316.64.0
Topology - topology: https://github.com/ranj063/soft-1/tree/sof-v0.55
DSP Firmware - https://github.com/ranj063/sof/tree/sof-v0.55

The test item "SpeakerDMic" is to loop from speaker to digital microphone. You can refer to the factory test script as below.
https://chromium.googlesource.com/chromiumos/platform/factory/+/master/py/test/pytests/audio_loop.py

In this test, it will run the audiofuntest to test the function, something like below:
audiofuntest -r 48000 -i hw:0,0 -o hw:0,0 -l 5 -a 0

INFO kernel: [  329.372343] sof-audio sof-audio:  Firmware info: version 1.1-13be5 build 32 on Dec 24 2018:19:43:24
INFO kernel: [  329.374091] sof-audio sof-audio: firmware boot complete
ERR kernel: [  336.048321] sof-audio sof-audio: error: ipc timed out for 0x60040000 size 0xc
ERR kernel: [  336.048355] sof-audio sof-audio: status: fw entered - code 00000005
ERR kernel: [  336.048495] sof-audio sof-audio: error: unexpected fault 0x00000000 trace 0x00004000
ERR kernel: [  336.048502] sof-audio sof-audio: error: waking up any trace sleepers
ERR kernel: [  336.048539]  Speakers: ASoC: trigger FE failed -110
ERR kernel: [  336.048864] sof-audio sof-audio: error: rx list empty but received 0x0
ERR kernel: [  336.048876] sof-audio sof-audio: error: can't find message header 0x60040000
ERR kernel: [  336.048935] sof-audio sof-audio: XRUN for unknown stream, msg_id 0
ERR kernel: [  336.051095] sof-audio sof-audio: XRUN for unknown stream, msg_id 0
ERR kernel: [  336.051218] sof-audio sof-audio: XRUN for unknown stream, msg_id 0
INFO goofy[1236]: Test main_fleex:SMT.SpeakerDMicManual.SpeakerDMic_16 (03e10db4-ec6c-4fdf-8e85-e04c0fe8ce68) completed: FAILED (Under output volume 10; Failed to parse audiofuntest output#012Traceback (most recent call last):

Activity

  1. ranj063
    assigned
    michalgrodzicki
    on Jan 8, 2019

  2. wenqingfu commented on Jan 9, 2019

    @wenqingfu
    wenqingfu
    on Jan 9, 2019

    same error as #743 ?

  3. xiulipan commented on Jan 9, 2019

    @xiulipan
    xiulipan
    on Jan 9, 2019
    Contributor

    @keyonjie

    ERR kernel: [  336.048321] sof-audio sof-audio: error: ipc timed out for 0x60040000 size 0xc
ERR kernel: [  336.048355] sof-audio sof-audio: status: fw entered - code 00000005
ERR kernel: [  336.048495] sof-audio sof-audio: error: unexpected fault 0x00000000 trace 0x00004000
ERR kernel: [  336.048502] sof-audio sof-audio: error: waking up any trace sleepers
ERR kernel: [  336.048539]  Speakers: ASoC: trigger FE failed -110
ERR kernel: [  336.048864] sof-audio sof-audio: error: rx list empty but received 0x0
ERR kernel: [  336.048876] sof-audio sof-audio: error: can't find message header 0x60040000

    It seems the IPC is handled, but not in time here.

    @naveen-manohar
    will try to reproduce it and have some feedback.

  4. keyonjie commented on Jan 9, 2019

    @keyonjie
    keyonjie
    on Jan 9, 2019
    Contributor

    we need to duplicate it on our side to root cause it, but simultaneous pipelines stress test fails in v0.55 is our known issue.
    Shall we align on that we will focus on fixing it in PV release, but not v0.55?

  5. keqiaozhang commented on Jan 9, 2019

    @keqiaozhang
    keqiaozhang
    on Jan 9, 2019
    Collaborator

    I used my test script to test dmic capture and speaker playback at same time, it passed 1000 iterations without errors.
    CMD:
    arecord -Dhw:0,99 -f dat -c 4 /dev/null & aplay -D hw:0,0 -f dat -c 2 /dev/zero & sleep 3; pkill aplay; pkill arecord

    I also tried to test it with audiofuntest, but failed with "Failed to exec client: No such file or directory"

    localhost /usr/local/factory # audiofuntest -P -r 48000 hw:0,0 -C 2 -R -r 48000 -c 4 hw:0,99 -l 5
Config values.
        Speaker active channels: 0, 1
        Mic active channels: 0, 1, 2, 3
        Allowed delay: 1.5000(s)
        FFT size: 2048
        Confidence threshold: 3.0000
        Match window size: 7
        Player parameter: -r
        Player FIFO name:
        Recorder parameter: -r
        Recorder FIFO name:
        Sample format: s16
        Sample rate: 64000
        Number of Microphone channels: 4
        Number of Speaker channels: 2
        Number of test rounds: 10
        Tone length: 5.0000(s)
        Volume gain: 50
Failed to exec client: No such file or directory
Failed to exec client: No such file or directory
Killed

    @naveen-manohar
    What have I missed? Could you share same docs about how to setup the env and run audiofuntest on DUT?

  6. mengdonglin
    added
    bugSomething isn't working as expected
    on Jan 9, 2019

  7. dabekjakub commented on Jan 9, 2019

    @dabekjakub
    dabekjakub
    on Jan 9, 2019
    Member

    I see that topology and FW are from @ranj063 repo.
    Does this reproduce on FW master branch ?
    @naveen-manohar

  8. ranj063 commented on Jan 9, 2019

    @ranj063
    ranj063
    on Jan 9, 2019
    Collaborator

    @dabekjakub this bug was reported with R72
    @naveen-manohar can you please check if this can be fixed in 1.0 and not for 0.55?

  9. naveen-manohar commented on Jan 10, 2019

    @naveen-manohar
    naveen-manohar
    on Jan 10, 2019
    ContributorAuthor

    @ranj063 Yes i'm tracking the same with ODMs
    Shall update soon they give a positive nod

  10. naveen-manohar commented on Jan 10, 2019

    @naveen-manohar
    naveen-manohar
    on Jan 10, 2019
    ContributorAuthor

    Got customer's reply - "SOF v1.0 will not be backported to the factory branch. We need to make sure this gets fixed on the factory branch."

  11. wenqingfu commented on Jan 10, 2019

    @wenqingfu
    wenqingfu
    on Jan 10, 2019

    @dabekjakub @keqiaozhang , you need to talk with @keyonjie if it's the same problem he run into in this comment . Before that, he reported 1000+ successful iterations of a similar test scenario as well.

    Good luck!

  12. keqiaozhang commented on Jan 11, 2019

    @keqiaozhang
    keqiaozhang
    on Jan 11, 2019
    Collaborator

    @naveen-manohar
    I can reproduce this issue with:
    audiofuntest -P "aplay -D hw:0,0 -r 48000 -f s16 -t raw -c 2 -B 0 -" -R "arecord -D hw:0,99 -r 48000 -f s16 -t raw -c 4 -B 0 -" -r 48000 -T 5 -a 0 -m "0,1" -c 4 -g 100
    Also, this is a known issue to us and track it in #743, we will fix it in PV release.
    This issue is easier to reproduce if three pipelines are running at same time.

  13. wenqingfu commented on Jan 11, 2019

    @wenqingfu
    wenqingfu
    on Jan 11, 2019

    This issue is easier to reproduce if three pipelines are running at same time.

    interesting finding, thanks! can you please put down the steps you used for creating "three pipelines"? What about more than three pipelines? Would it make the issue reproduction even easier?

  14. wenqingfu commented on Jan 11, 2019

    @wenqingfu
    wenqingfu
    on Jan 11, 2019

    scouted symptom report in #784 shows similar error pattern, too.

  15. mengdonglin
    added
    GLKApplies to Gemini Lake platform
    on Jan 14, 2019

  16. 5 remaining items

  17. keyonjie commented on Jan 16, 2019

    @keyonjie
    keyonjie
    on Jan 16, 2019
    Contributor

    Hi @keyonjie,
    We tried the fix shared by you for 50 iterations and wasn't able to hit the issue with factory test - SMT.SpeakerDMic
    But Basic Audio Sanity Failed, Both Speaker & Headset playback observes bursting noise (more frequently observed during audio playback soon after boot)

    @naveen-manohar can you confirm if this noise issue is regression after apply my new fixes?

  18. naveen-manohar commented on Jan 16, 2019

    @naveen-manohar
    naveen-manohar
    on Jan 16, 2019
    ContributorAuthor

    @Keyon yes we did re-verify and bursting/crackling issue is reproduced only the firmware change.
    Without the firmware change, default cpfe works fine.

  19. keyonjie commented on Jan 16, 2019

    @keyonjie
    keyonjie
    on Jan 16, 2019
    Contributor

    @Keyon yes we did re-verify and bursting/crackling issue is reproduced only the firmware change.
    Without the firmware change, default cpfe works fine.

    OK thanks for the information, I figured out the fixes in UP^2 with nocodec, so let me try to figure out which patch is the convict, I suspect this one at my first think:

    pipeline: switch schedule_task_idle to schedule_task

    can you do a quick try if removing this commit make difference?

    I will do debugging here tomorrow.

  20. keqiaozhang commented on Jan 17, 2019

    @keqiaozhang
    keqiaozhang
    on Jan 17, 2019
    Collaborator

    @naveen-manohar
    I did not observe the bursting noise you mentioned, please share your test recipe or better provide us the noise wav file.

  21. naveen-manohar commented on Jan 17, 2019

    @naveen-manohar
    naveen-manohar
    on Jan 17, 2019
    ContributorAuthor

    @keqiaozhang Steps we tried:

    1. Boot to R72-11316.75.0 OS & replaced with dsp firmware /lib/firmware/intel/sof-glk.ri with below md5sum and reboot - edd33e76454176f45151fc842ad2c7d7
    2. Observation: Speaker audio playback to notice bursting noise soon after boot

  22. keqiaozhang commented on Jan 17, 2019

    @keqiaozhang
    keqiaozhang
    on Jan 17, 2019
    Collaborator

    @naveen-manohar
    Now I can reproduce this bursting noise issue, but this issue only happens with browser, no such noise with aplay.
    After bisection, we found that this noise is caused by Keyon's new fix(debug: hda-dma: set timeout for preload). Now he is working on it.

  23. keyonjie
    assigned
    mmaka1
    and
    slawblauciak
    on Jan 17, 2019

  24. keyonjie commented on Jan 17, 2019

    @keyonjie
    keyonjie
    on Jan 17, 2019
    Contributor

    hi all, the noise issue is introduced by my try to workaround hda-dma preload timeout, it workaround the wait timeout issue, but noise happens at this DGCS_BF fail to become true cases.

    I have just created a new issue #856 to track that.

    @mmaka1 @slawblauciak can you please help on that?

  25. thesofproject
    deleted a comment from naveen-manohar on Jan 17, 2019
  26. michalgrodzicki
    assigned
    keyonjie
    on Jan 28, 2019

  27. wenqingfu commented on Jan 29, 2019

    @wenqingfu
    wenqingfu
    on Jan 29, 2019

    Can this bug be closed because of c3111ec?

  28. keqiaozhang commented on Jan 30, 2019

    @keqiaozhang
    keqiaozhang
    on Jan 30, 2019
    Collaborator

    I cannot reproduce this issue any more.
    @naveen-manohar
    Please help to close this issue if you are satisfied with it. Thanks!

  29. naveen-manohar commented on Jan 30, 2019

    @naveen-manohar
    naveen-manohar
    on Jan 30, 2019
    ContributorAuthor

    Thanks a Ton, fix has been validated & we can close this bug

  30. naveen-manohar
    closed this as completedon Jan 30, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

GLKApplies to Gemini Lake platformP1Blocker bugs or important featuresbugSomething isn't working as expectedduplicateThis issue or pull request already exists

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions