Happened again just now. This is unacceptably flaky behavior for a consumer product. Who can triage this and tell me what logs to gather to diagnose?

Hi kbr@, thanks for filing the bug.
Could you please send a feedback report with alt+shift+i ?
To get proper debug information, the feedback report should be generated at the time while headphone is plugged but not detected.

You can mention the bug number 898290 in the feedback report so we can locate it.

Thanks!

It just happened again and I've submitted feedback via this method, referring to this bug. It seems to happen after multiple suspends/resumes of the laptop. I just rebooted yesterday to fix this issue and it's already happening again.

I've seen this periodically on eve over the period of many months as well. I've filed a feedback report at https://listnr.corp.google.com/report/85809935389.

Unfortunately, there is no suspicious log in #4.
Headphone and mic are not detected.
No error message from rt5663 headset codec.

Based on observation in #3, we setup a test on eve to run
suspend_stress_test
and check audio jack status.
Once the issue is reproduced, we can dump the register map and check whether there is any register not in the right place.

We will see if we can have more solid repro step, and ask for realtek's help for debugging.

Just happened again and submitted another feedback report via alt-shift-i. The machine was rebooted as recently as yesterday, so it should be easy to manually reproduce this - just suspend/resume the machine a few times and try plugging in and unplugging the headphones. (Mine are a headset with integrated microphone.)

I can't reproduce it even if I suspend and resume for more than 1000 times. Do I need to play music when I suspend? Or do you have any other suggestion to trigger it? Thanks.

All I did the last time was use my machine normally, without headphones, for several hours - including several suspends/resumes - and then when I plugged in the headset, it wasn't detected.

Is there any more detailed logging I can provide from my system? Could some daemon process be dying, and would that be evidenced anywhere? Can I probe the state of registers on any chip that controls the audio input?

You can try it to read regmap (Need root permission)

echo 1> /sys/kernel/debug/regmap/i2c-10EC5663\:00/cache_bypass  (do not use cache)
cat /sys/kernel/debug/regmap/i2c-10EC5663\:00/registers

Could you give a pointer to instructions on how exactly to set up my system to be able to run these commands? Enabling developer mode, opening the alternate console, etc.? Thanks.

P.S. I have the Crostini environment installed but obviously these registers aren't exposed to that environment.

Yes, you need to enable developer mode to get root permission.
1. Recovery mode: Esc + Reload held while pressing Power
2. Developer mode: Enter recovery mode and press ctrl+d
But this movement will clean all your data.

After entering developer mode, you can type ctrl+alt+t on browser to open terminal. Then type shell to get shell. Switch root by "sudo -i" or "sudo su". Password is "test0000". Thanks.

Sigh. Is there any way to back up and restore the Crostini environment so I don't have to rebuild this machine from scratch?