Skip to main content ↑ ↓
Topic: Complete system freeze (Read 1155 times) previous topic - next topic
0 Members and 2 Guests are viewing this topic.

Complete system freeze

For about a week or two (can't quite remember exactly when it started) my system has started randomly freezing once every day or two: completely unresponsive to anything other than a hard reboot, and the system gets very hot with the fans running full speed. It mostly occurs immediately prior to entering sleep, but also rarely (as in once, today) froze when just sitting there not doing anything.  /var/log/kernel.log is filled with something akin to below, repeated hundreds of times per second—in a freshly-rotated kernel.log, it grew to 500 MiB over the course of about 5–10 minutes from the system freeze to when I noticed it was frozen. This was taken from the last set of logspam before I hard rebooted the system, so it seems likely to me that the “No installed handler for fixed event - SleepButton” message was from me holding the power button before the hard shutdown circuitry kicked in.

I'm on a 2015 MacBook Pro with OpenRC and using KDE. The issue occurs on both linux (6.5.5.artix1-1) and linux-lts (6.1.54-1), but notably did not occur on linux-lts until recently, it started happening on both kernels simultaneously.

Code: [Select]
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: No installed handler for fixed event - PM_Timer (0), disabling (20230331/evevent-255)
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: No installed handler for fixed event - SleepButton (3), disabling (20230331/evevent-255)
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: Could not disable RealTimeClock events (20230331/evxfevnt-243)
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: No handler or method for GPE 00, disabling event (20230331/evgpe-839)
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: No handler or method for GPE 01, disabling event (20230331/evgpe-839)
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: No handler or method for GPE 02, disabling event (20230331/evgpe-839)
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: No handler or method for GPE 03, disabling event (20230331/evgpe-839)
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: No handler or method for GPE 04, disabling event (20230331/evgpe-839)
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: No handler or method for GPE 05, disabling event (20230331/evgpe-839)
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: No handler or method for GPE 11, disabling event (20230331/evgpe-839)
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: No handler or method for GPE 12, disabling event (20230331/evgpe-839)
Oct  3 18:09:05 GLaDOS kernel: ACPI Error: No handler or method for GPE 13, disabling event (20230331/evgpe-839)

Re: Complete system freeze

Reply #1 –

 It does same error when you boot from a live USB? You may need to clean the fans if you haven't done it for a long time. If it gives same error while on live environment it's a hardware issue.



Re: Complete system freeze

Reply #2 –
Quote
but notably did not occur on linux-lts until recently
In the galaxy repo there is linux-lts515, which stays clear from the screwups of the 6.x series, before you can get more elaborate help try with that.  :)

Re: Complete system freeze

Reply #3 –
If you go back to 5.15 you would also need to install linux-firmware-xz:
https://forum.artixlinux.org/index.php/topic,5762.msg35984.html#msg35984
You can get older kernels (and related headers if needed) from the Artix archive here to find when the problem started, if it was due to the kernel:
https://archive.artixlinux.org/packages/l/linux/
https://archive.artixlinux.org/packages/l/linux-headers/
(My M4500 Nvidia / nouveau has also started freezing / crashing today, although such things are not unheard of with this graphics combination it has suddenly got much much worse. I just downgraded mesa, lib32-mesa and cairomm as the most likely recently upgraded candidates, but don't yet know if that has worked. I wondered if it was the new version of the  Waterfox browser I was using, but I downgraded that and it still froze up again afterwards.)

Re: Complete system freeze

Reply #4 –
Quote from: Surf3r – on
It does same error when you boot from a live USB? You may need to clean the fans if you haven't done it for a long time. If it gives same error while on live environment it's a hardware issue.
It's an 8 year old laptop but I still do basic maintenance. It doesn't run hot at all until it freezes, which I assume is because whatever part of the kernel is generating all the ACPI logs is just spinning and consuming 100% of some/every CPU core.

Later today when I have some time I'll make a fresh live USB with the latest Plasma graphical installer and experiment with reproducing it with that.

Quote from: Hitman – on
Quote
but notably did not occur on linux-lts until recently
In the galaxy repo there is linux-lts515, which stays clear from the screwups of the 6.x series, before you can get more elaborate help try with that.  :)
I downgraded linux and linux-headers to 6.5.2.artix1-1 from 6.5.5-artix1-1 and linux-lts and linux-lts-headers to 6.1.51-1 from 6.1.54-1 and I haven't been able to reproduce it, so I guess it has to be the kernel (I was thrown off by it also happening on LTS, maybe the bad change was backported). If it starts happening again with these I'll experiment with downgrading all the way to 5.15. I might also look into doing a bisect to figure out exactly what is causing it if it doesn't get fixed in the next Kernel update.
 

Re: Complete system freeze

Reply #5 –
Downgrading those things I mentioned fixed my freezes, after upgrading cairomm again it's fine so far. It looks like mesa-1:23.2.1-1-x86_64.pkg.tar.zst is the problem.
There is this bug report related to gnome-shell crashing caused by that new version of mesa:
https://bugs.archlinux.org/task/79831?project=1&string=mesa
But then on the Arch forum there is this thread about gnome-shell which apparently resolves the same issue by kernel downgrades or patches to the AMD gpu driver:
https://bbs.archlinux.org/viewtopic.php?id=288735
So could it be some combined compatibility problem between the kernel and mesa?

Re: Complete system freeze

Reply #6 –
Quote from: ####### – on
Downgrading those things I mentioned fixed my freezes, after upgrading cairomm again it's fine so far. It looks like mesa-1:23.2.1-1-x86_64.pkg.tar.zst is the problem.
<snip>
So could it be some combined compatibility problem between the kernel and mesa?
Hmm, according to pacman.log, mesa and lib32-mesa were upgraded around when I started having issues. So I tried re-upgrading all the Kernel-related packages and then downgrading mesa & lib32-mesa to 1:23.1.8-1 from 1:23.2.1-1 and cairomm to 1.14.4-1 from 1.14.5-1; but unfortunately I still had my freezes. I suspect that since my errors seem to be related to ACPI GPE interrupts, it's a separate issue from yours.

I'll keep the Linux packages ignored for now until the next Kernel upgrade, since it's the lowest-effort fix I've found lol.

Re: Complete system freeze

Reply #7 –
That sounds like it must be the case then, I updated my E7470 with Intel graphics and that seems fine with no crashes, at least something here works properly.  ;D