Problem description: Linux suffers random crashes especially during video intensive operations.
Hardware configurations:
CPU: AMD Ryzen 7 5700G with Radeon Graphics
Motherboard: Gigabyte B550M D3H
BIOS version: F14 then F15c, both exhibited the same behaviors. F15c seems to be a little more resilient to the crashes, meaning the execution of glmark2 lasts a little longer before crashes.
OS: Linux Mint 20.3 / with a number of Ubuntu kernels from 5.11, 5.13, and 5.15. Due to the CPU and motherboard, have to use non-packaged kernels for both graphics and network interface to work.
Key findings:
Once the CPU virtualization is turned-on, system starts to crash randomly. For Gigabyte motherboard, Hardware virtualization is called SVM mode. All crashes occurred under CPU SVM mode is enabled.
Crashes means, sudden lost of desktop screen during operations and received only a black screen. No keyboard or mouse response. Had to power cycle the computer to resume operations. Tried to store the syslog for each boot ended in crash.
At beginning, I tried to sync the file system at rebooting the computer, but the computer boots up without syncing the file system as well.
During at least three months of this kind of bad experiences, tried to upgrade kernel from 5.11 to 5.15, none of them helped.
Methods of replicating the crashes:
1. Turn on the SVM mode in BIOS.
2. Execute glmark2 to benchmark the graphic card performance.
3. System crashes before the benchmark test is done every time.
To me the syslogs doesn't show any meaningful info for the crashes. The following link gives some of the syslog gathered in the painful experiences especially those crashes shortly after reboot.
https://drive.google.com/drive/folders/1cYbQPdfPSC_6A879xBWRs1jIY7R5ylMU?usp=sharing
The most important Finding is once the SVM mode on the motherboard is turned off (disabled), I no longer experience the same kind of random crashes, and glmark2 benchmark can finish without any issues. All versions of kernel 5.11 to 5.15 work stably as expected.
The real question to me is how the CPU virtualization is supported in the kernel and why turning on the CPU virtualization can promote the crashes. Perhaps the BIOS might be involved for a satisfactory solution to this observed problem, however, what I can offer is just a sure way to replicate the crashes for others to investigate this issue further.
That's all I can report here. If you need any further information. Please feel free to let me know.
Thanks for your attentions.
This bug is missing log files that will aid in diagnosing the problem. While running an Ubuntu kernel (not a mainline or third-party kernel) please enter the following command in a terminal window:
apport-collect 1982965
and then change the status of the bug to 'Confirmed'.
If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.
This change has been made by an automated script, maintained by the Ubuntu Kernel Team.