Dear Debian users,
Recently I installed Debian on an ASUS UX501J laptop. Debian stable (Buster)
works flawlessly (I'm using that now), but when I tried to upgrade to testing,
things started going wrong. I would get random kernel panics and other errors
on boot. Thinking I may have done something wrong during upgrade, I created a
fresh testing installation image on a USB stick and installed that.
Installation finished without errors, but I got the same problems on boot, even
the first one after install. The errors were the same (randomly, of course)
when booting in rescue mode. I took pictures of the screen (my apologies to
those using screen readers): see [2, 3, 4, 5].
The odd thing is that about every 1 in 5 boot sequences would succeed, and I
would be able to use KDE without issues. When I ran into trouble, I would use
the Alt + SysRq + (R E I S U B) sequence if possible, but after a kernel panic
the only option I saw was to use the power button. After enough unsuccessful
boot attempts, I would see filesystem corruptions (caused by the hard boot
after panics?) and I would be unable to boot entirely (fsck always segfaulted).
At some point I even ended up in an initramfs shell I never even knew existed.
How can I debug this problem? My suspicion is that this has to do with the
kernel upgrade between stable and testing (4.19 to 5.10), but I'm not sure.
Things I have tried so far:
- Ran smartctl test on the SSD: no issues reported.
- Ran memtest86: no issues reported [6].
(I am not suspecting hardware issues anyway, because buster is running fine
currently.)
- Chose different options for partitioning during install: tried encrypted
LVM, LVM and plain old ext4. Using full disk every time. This is because I
also saw error messages about LVM not being able to find the volume groups
a couple of times, but I did not take pictures unfortunately. It also
seemed to work anyway, because the partitions would be mounted correctly in
the few instances that booting succeeded.
- Installed non-free firmware (used the unofficial image [1]) during
installation and installed the intel-microcode package after installation
in an attempt to fix the error displayed in [4] ("[Firmware Bug]:
TSC_DEADLINE disabled due to Errata: please update microcode to
version: 0x22 (or later)").
As I wrote before, the unofficial image with non-free firmware for Buster
installed without issues and is what I am using currently. Still I would like
to figure out what is going wrong because (1) I normally like to use testing or
even unstable and (2) one day I'll have to upgrade this laptop anyway.
Thanks in advance, best regards,
Thom
[1]:
https://cdimage.debian.org/images/unofficial/non-free/images-including-firmware/
bullseye_di_rc1+nonfree/amd64/iso-cd/
[2]: https://cdn.pixabay.com/photo/2021/06/08/09/50/09-50-12-676_1280.jpg
[3]: https://cdn.pixabay.com/photo/2021/06/08/09/50/09-50-12-588_1280.jpg
[4]: https://cdn.pixabay.com/photo/2021/06/08/09/50/09-50-08-249_1280.jpg
[5]: https://cdn.pixabay.com/photo/2021/06/08/09/50/09-50-08-446_1280.jpg
[6]: https://cdn.pixabay.com/photo/2021/06/08/09/50/09-50-16-763_1280.jpg
