Additionally, F9 and F10 keys (usually control volume up and down) do not function anymore. 

Device still appears to Linux, but not to ChromeOS

Deleted: Screenshot 2016-06-03 at 4.35.47 PM.png 52.4 KB

	Screenshot 2016-06-03 at 4.35.47 PM.png 52.4 KB View Download

https://feedback.corp.google.com/#/Report/9993230331

Description:
I lost audio after updating to latest chrome for chromebox.

Version 51.0.2704.79 (64-bit)
Platform 8172.47.0 (Official Build) stable-channel guado
Firmware Google_Guado.6301.108.4

FWIW, I switched from Beta to Dev channel and the audio device came back. I did this as a last resort after all other attempts to bring it back failed: 1) rebooting 2) disconnecting / reconnecting audio cables 3) playing with all sound/volume settings in ChromeOS 4) restoring all chrome://flags back to default. 

This(guado) is a chromebox board with no internal speaker and microphone.
Internal output and input audio nodes are not expected in Audio Settings menu.

Did you mean you have externally plugged audio peripheral, for which you had audio nodes observed? Did you loose the HDMI audio?

I had an external headset (microphone and headphones) plugged into the audio jack on this box and when they are present the Audio Settings would previously show nodes nodes. Now the nodes are missing and no input or output via that jack occurs. A USB webcam/mike and a set of USB speakers both do show and still work.

Hi Kalin, can you reproduce the issue ?
I don't have a guado here, so I only checked the guado in the lab remotely. The  headphone node is enumerated by CRAS, but I am not sure if it can detect headphone correctly.
I requested a guado from chromestop here, but maybe need a few days to get one.
If you can reproduce the issue, could you please connect one device to network ? Maybe I can debug it remotely.
Thanks!

@Comment6 : I was referring to the headphones I had plugged in to the 1/8'' audio out jack on the Chromebox. When this happened, I was able to still see a microphone for an external usb webcam. However, since audio output stopped working, I'm not sure if that input worked at the time either.

OK, now we know this is about 3.5mm audio jack audio. 
I'll follow up with observations and logs tomorrow. 

helenzhang@ helped to reproduce. Reproduced 100%. After recovery with the stable build, the audio jack audio node is missing. HDMI audio too. Only USB audio node comes up.

localhost / # cras_test_client --dump_server_info 
System Volume (0-100): 100 
Capture Gain (0.00 - 23.81): 20.00dB 
User muted: Not muted
Output Devices:
        ID      Name
Output Nodes:
        Stable Id        ID      Vol   Plugged  L/R swapped           Time      Type             Name
Input Devices:
        ID      Name
        4       Post DSP Loopback
        3       Post Mix Pre DSP Loopback
Input Nodes:
        Stable Id        ID     Gain   Plugged  L/R swapped           Time      Type             Name
        (8c74f766)      4:0         0      yes              no           0      POST_DSP_LOOPBACK Post DSP Loopback
        (8a04af91)      3:0         0      yes              no           0      POST_MIX_LOOPBACK Post Mix Pre DSP Loopback
Attached clients:
        ID      pid     uid
        9       19232   1000
        11      24084   0
Num active streams: 1
Last audio active time: 841, 385240155

Generated log is at https://pantheon.corp.google.com/storage/browser/chromiumos-test-logs/bugfiles/cr/617273/

FYI next beta update (RC 8350.31.0 / 52.0.2743.41) does not reproduce the issue.

aplay -l and arecord -l show nothing.
Sound card is not detected at all.
This seems to be issue in kernel not cras.
I will check what is changed on R51.

Deleted: audio_diagnostics.txt 494 KB

audio_diagnostics.txt 494 KB View Download

Hi Kalin,
could you please plug a guado to network so I can debug remotely?
It will take a few days for me to get a guado from chromestop here. Thanks!

I installed 8172.55.0 on chromeos1-row3-rack5-host3, and sound card can be detected.
I then installed 8172.47.0, and sound card is missing.
I also got suspicious error messages like this:

[    7.949030] snd_hda_codec_via: Unknown symbol cleanup_module (err 0)
[    7.949063] snd_hda_codec_via: Unknown symbol init_module (err 0)
[    7.949332] snd_hda_codec_via: Unknown symbol cleanup_module (err 0)
[    7.949357] snd_hda_codec_via: Unknown symbol init_module (err 0)
[    7.951357] snd_hda_codec_via: Unknown symbol cleanup_module (err 0)
[    7.951398] snd_hda_codec_via: Unknown symbol init_module (err 0)
[    7.951701] snd_hda_codec_via: Unknown symbol cleanup_module (err 0)
[    7.951732] snd_hda_codec_via: Unknown symbol init_module (err 0)

This happens when snd_hda_* module probe fails.

I tried reboot three times and the issue is consistent.
However after that, autotest server takes over control of it and repair it into R53.

At the same time, I found that chromeos1-row4-rack5-host3 was on 8172.47.0, and sound card was missing too.

I think there is definitely something wrong on R51.
But diff between 8172.55.0 and 8172.47.0 does not look suspicious https://crosland.corp.google.com/log/8172.47.0..8172.55.0.

I also checked the diff in kernel under sound/ between R50 and R51
git log cros/release-R51-8172.B-chromeos-3.14..cros/release-R50-7978.B-chromeos-3.14 sound/
I shows only two unrelated changes in ASoC.

I tested on 8172.55.0 and 8172.47.0 more times.
On 8172.47.0, it failed to detect sound card on 5/5 reboots.
On 8172.55.0, it can detect sound card on 5/5 reboots.
I can do more testing, but it seems that there is something in this diff list
https://crosland.corp.google.com/log/8172.47.0..8172.55.0
fixed the issue.
This is the only CL that is related to guado:
https://chrome-internal-review.googlesource.com/#/c/262275/
It reverted https://chrome-internal-review.googlesource.com/#/c/258676/

But I have no idea how would that affects audio.
Note that https://chrome-internal-review.googlesource.com/#/c/258676/ landed in 8172.29.0, so I can do more testing around that version to confirm it is the CL to blame.

#CBC-RS/TC-watchlist

Update some test result:

I tested various image versions on R51-8172.X.0:

28 good
31 good
40 good
44 good
45 good
47 bad
54 good
55 good

It is weird that 8172.47.0 is the only version that is bad.
The build failed on 46 and 48~53 so we only have 45, 47, 54.
I also removed ui.conf to disable ui. The issue persiste on 47, so we don't need to look into chrome.

45 -> 47: some CL broke sound card detection.
https://crosland.corp.google.com/log/8172.45.0..8172.47.0

Nothing really related to guado.
There were two CLs on kerne 3.14
https://chromium-review.googlesource.com/#/c/348322/
https://chromium-review.googlesource.com/#/c/348323/
But those were for ARM and rockchip.

47->54: some CL fixed sound card detection.
https://crosland.corp.google.com/log/8172.47.0..8172.54.0
One CL related to guado was https://chrome-internal-review.googlesource.com/#/c/262275/
But I don't think it is related because https://chrome-internal-review.googlesource.com/#/c/258676/ was introduced from 29, and 31 was good.

Now, since 8172.47.0 is the only version that can not detect sound card, user will not hit by this issue once they get 8172.54.0 or later.
Still we need to figure out what is wrong with sound card detection. The error message in #16 is suspicious. Whenever snd_hda_* driver load fails, it shows that error message. I think we can create an issue on partner issue tracker and let guado-soc investigate further and find the root cause.

Created https://code.google.com/p/chrome-os-partner/issues/detail?id=54491 to track the root cause there.

Hi Kalin, could you please test with a guado and verify that this actually works ? I checked remotely on chromeos1-row4-rack5-host3 ,
reboot, and check aplay -l shows sound card, repeat for 10 times. I did not test audio jack detection but I think it should work as long as sound card is there.
Once you verify it, we can close this issue as Wontfix since image version later than 8172.54.0 works fine.
Thanks!

Tested with Guado on R51-8172.55.0, and all is working fine. Audio jack and external display audio is working as intended.

Hi All, We have a Chromebox for Meetings that has this issue on the latest stable release. Checking the version numbers, it is on 8172.47, picture attached. 

Do we have an eta for next update date to address this?

Deleted: 2016-06-16.jpg 362 KB

	2016-06-16.jpg 362 KB View Download

Hi Bernie, do you have an eta for the next update on stable channel?
It might take some time to qualify new stable version for all boards. Is it possible to push 8172.55.0 for guado only?
Thanks!

Ideally the snd-hda-codec-via should not be loaded, it looks to me kina binaries error regarding snd-* modules.

# modinfo snd-hda-codec-via.ko                                                                                                         
filename:       /lib/modules/3.14.0/kernel/sound/pci/hda/snd-hda-codec-via.ko
description:    VIA HD-audio codec
license:        GPL
alias:          snd-hda-codec-id:1106*
depends:        snd-hda-codec,snd-hda-codec-generic
intree:         Y
vermagic:       3.14.0 SMP preempt mod_unload

Hi Harry,
The snd-hda-codec-via error message occurs when I tried modprobe snd-hda-intel on 8172.47.0. I am not sure why snd-hda-codec-via is loaded though.
I attached the snd-hda-* modules on 8172.47.0 and 8172.55.0.
Will this help you check what is wrong?
Thanks!

Deleted: snd-hda-modules.tgz 268 KB

snd-hda-modules.tgz 268 KB Download

We are testing 8172.56 today, it could start rolling out (low %) as early as this afternoon, if not then Monday, assuming nothing bad is discovered in testing.

We are verifying this on guado with hotrod as well. Please wait for a go-ahead from me before pushing 8172.56. Will update the bug soon.

Headphone jack and HDMI audio work fine on hotrod enrolled guado with 8172.56.0. Good to go from our end.

Audio Jack works in regular consumer mode too for 8172.56.0 build - next stable RC.

Sounds like fixed to me, though I am still curious why one build was bad (corrupted artifacts or something?). 

re #25/#31, I did triage on chrome-os-partner:54491 

It looks to me the key file of snd-hda-codec.ko w/ incorrect symbols breaks audio subsystem modules loading, while I haven't dived into repositories like Cheng-Yi did before.

 Issue 619444  has been merged into this issue.

Ongoing reports from users on CBC
#CBC-RS/TC-watchlist