You are not logged in.
system information
Current installed kernel:6.18.49-2-lts
CPU:AMD Ryzen 7 6800H
GPU:AMD Radeon 680M integrated graphics
Laptop:Lenovo ThinkBook 16+
BIOS:J6CN50WW
BIOS date:09/27/2024
Before testing Linux, I reset all BIOS settings to factory defaults, except for the Secure Boot setting, which I left manually configured.
There is no CPU overclocking, Curve Optimizer, PBO tuning, custom UMA configuration, or other performance-related BIOS modification currently applied.
Problem
The machine randomly reboots when running Linux.
The problem already happens inside the Arch Linux live environment, before the OS is installed.
Normally the live environment survives for somewhere around 10–20 minutes and then suddenly reboots. Sometimes it lasts longer or shorter, so the interval is not completely fixed.
I have seen this happen while the system was idle as well as while I was actively installing packages.
The BIOS setup screen itself is stable. I left the machine sitting inside the BIOS for about 30 minutes and it did not reboot.
I have also used Windows on this laptop before. Windows sometimes BSODed during the boot process, before successfully reaching the desktop, but once Windows successfully booted, I do not remember it ever randomly rebooting by itself during normal use.
This makes me unsure whether this is a Linux power-management/kernel issue, a firmware issue, or some underlying CPU/RAM/platform stability problem.
I should also mention that I am still a newbie when it comes to Linux. Some of the troubleshooting ideas and the way I categorized the possible causes were put together with the help of AI, so if any of my terminology, assumptions, or troubleshooting direction is incorrect, please feel free to point it out.
Kernel panic
During one of the Linux crashes I managed to see an actual kernel panic on screen.
The important message was:
Kernel panic - not syncing: Attempted to kill the idle task!Most of the other reboots happen too quickly for me to capture the panic screen.
After one reboot, Linux reported:
x86/amd: Previous system reset reason [0x00080800]: software wrote 0x6 to reset control register 0xCF9I did not see obvious MCE / Machine Check / Data Fabric / thermal reset errors in the kernel log.
I only saw normal initialization messages such as:
MCE: In-kernel MCE decoding enabled.RAS: Correctable Errors collector initialized.&
amdgpu: RAS: optional ras ta ucode is not availableKernels / environments tested
The issue happened in the Arch live environment repeatedly.
I also tried an older Arch installation ISO instead of only the latest one, and the machine still rebooted.
Because the live environment was unstable, I installed the system as quickly as possible and currently have:
Linux 6.18.49-2-lts
The installed LTS system still randomly reboots as well.
Therefore this does not appear to be limited to only one Arch live ISO.
CPU frequency / power-management observations
The normal CPU frequency driver was:
amd-pstate-epp
At first I suspected AC power because the laptop seemed to reboot more frequently when plugged in.
However, I later tested it on a fully charged battery with the AC adapter disconnected, and it could still reboot very quickly.
CPU frequencies also appeared fairly similar between AC and battery operation, so AC power itself no longer seems to be a necessary condition for triggering the issue.
Kernel parameters already tested
I tested the following parameters individually from GRUB by pressing e and appending them to the kernel command line:
1. nomodeset
This made things worse. The machine basically rebooted immediately or could not boot normally, so I stopped testing it.
2. idle=nomwait
The system booted, but eventually rebooted again. No fix.
3. amd_pstate=passive
This was interesting because on at least one test the machine remained stable noticeably longer than usual. However, later tests still rebooted, so it is not a reliable fix.
4. amd_pstate=disable
The machine rebooted very quickly. I did not have enough time to conclusively verify that it had successfully fallen back to acpi-cpufreq.
5. processor.max_cstate=1
Still rebooted.
6. cpuidle.off=1
Still rebooted.
I also used systemd-inhibit to block normal sleep/idle/shutdown actions. The inhibitor was confirmed active with systemd-inhibit --list, but the machine still rebooted.
Therefore this does not appear to be a normal systemd idle/suspend action.
Other things already ruled out or tested
I initially used USB tethering from a phone and saw several:
rndis_host ... NETDEV WATCHDOG: transmit queue 0 timed out
messages.
However, I completely disconnected the phone/USB tethering and the laptop still rebooted, so RNDIS does not appear to be the root cause.
I also enabled ZRAM later, but the reboot still occurred, so lack of swap/ZRAM was not the cause.
The Btrfs installation itself does not appear related either, because the machine was already rebooting inside the Arch live environment before installation.
Interesting behavior
The most interesting observations so far are:
The BIOS screen can remain stable for at least ~30 minutes.
Linux can reboot within roughly 10–20 minutes, sometimes much sooner.
Windows used to sometimes BSOD before reaching the desktop, but after successfully reaching the desktop it was generally stable.
amd_pstate=passive appeared to increase the stable runtime during at least one test, but did not permanently solve the problem.
Completely disabling cpuidle with cpuidle.off=1 did not solve it.
The AMD reset-reason message reports a software-triggered reset through 0xCF9.
What I would like help determining
I am not sure what subsystem is responsible for this issue.
I would appreciate any suggestions on what logs, kernel parameters, firmware information, or hardware tests would be most useful to narrow it down.
If anyone has seen similar random reboots or the specific:
Kernel panic - not syncing: Attempted to kill the idle task!
on a Ryzen 7 6800H system, I would be very interested to know what the cause was and how it was diagnosed or fixed.
Offline
Offline