Need help: Chromebook (Krane) broken? No network, no sound, no clock any more!

Hello,

as described here, I tried to make my touch screen rotate at sddm's login screen. I followed the advice given by DvaMishkiLapa, but I must have made a huge mistake and now my CB seems broken. I had to do several cold bootings (holding power button until CB shuts off), and after some attempts PMOS finally startet and I could log in. I then managed to undo the changes to /usr/share/sddm/scripts/Xsetup.

BUT neither the wifi device nor the sound device are found any longer, the system tells me there are no such devices. Furthermore, the system clock is set to January 1st, 1970, every time I start the CB. I can easily set the system clock by doing "date --set='2024-09-19 22:10', but when I try to use the command "hwclock" to set the hardware clock, the program tells me that no usable rtc can be found.

I really do need help there: would could have done wrong, that even the RTC is no accessible any longer?

How do I repair the system, without having to do a complete installation again, that would wipe my data?

Many, many thanks in advance for any help!!

added chromeos type::support labels

The directory that is supposed to contain all the modules for the running kernel was completely empty!

I got suspicions when lsmod showed only 4 or 5 modules loaded. Then I looked into /lib/modules/6.6.25-mt8183/ to search for the module that is needed for the rtc. Nothing to see there!

Then I manually unpacked the kernel package from /var/cache/apk and copied the modules to the place they belong and after the next reboot the system was working as expected again: the clock was running and the wifi device was working.

The big question is: what incident deleted all those modules? I have absolutely no idea! I never touched the modules dir before.

Most likely it is pure coincident that this happened at the same time I tried to make the touch screen rotate. But did this ever happen to anyone else? Does anyone have an idea what could have caused this?

By Wolfgang on 2024-09-20T18:22:20

Sounds like #3186 ?

By Hacker1245 on 2024-09-20T18:22:20

Very possible. I think I might have a slight clue of what could have happened:

In /var/cache/apk there are two kernel packages lying around: 6.6.25 and 6.6.36. When the mistake happened, 6.6.25 was the active kernel. (uname -a). I remember that one or two days earlier Discover told me there were updates available, which I installed. That way the kernel package 6.6.36 must have appeared at my Duet's hard drive.

BUT I did not reboot the device after installing the updates!

Next thing that happened to my Duet was me, turning it of the hard way by holding down the power button, because X wouldn't start due to the changes I had made, and there was no way of rebooting the soft way. Is it possible, that the upgrading process did somehow empty /lib/modules/6.6.25-mt8183/ already, but the installation of kernel 6.6.36 was interrupted when the initram was build? So, when I rebooted the device, 6.6.25 was still marked as the kernel to boot, but its modules were already gone.

To me, this makes sense, but I am not an expert on this issue.

By Wolfgang on 2024-09-21T10:34:46

@WeKa did you try doas hwclock --set..?

By Robert Eckelmann on 2024-09-23T17:44:57

Yes, but it didn't work, because the necessary module was not available.

By Wolfgang on 2024-09-23T17:44:57

@WeKa Just did a new install on a google-krane in the last days, and beside running into an issue with gnome-shell I did not experience the same issues as you mentioned before.

I used a USB-memory device with the pre-build image flashed to it for the first boots, and then ran pmbootstrap on the device to build its own image.

By Robert Eckelmann on 2024-10-01T14:07:51