Random System-Wide Freezes - AMDGPU (Ryzen 3 3200G / Vega)

[zedxot]

Hello,

My system is freezing completely, requiring a hard power off and reboot each time. I am seeking advice on how to diagnose and resolve this problem.

System Configuration:
Kernel Version: 6.12.16-gentoo-x86_64 (issue has been present on other recent kernels as well)
Processor (CPU): AMD Ryzen 3 3200G with Radeon Vega Graphics (Picasso APU)
Integrated Graphics (GPU): Radeon Vega Graphics (integrated in Ryzen 3200G)
Motherboard: MSI B450M PRO-M2 V2 (MS-7B84)
BIOS Version: 4.F3 (latest version from MSI)
Memory (RAM): 12GB DDR4 3200MHz (8GB + 4GB)
Window Manager: DWM

Description of the Problem: System-Wide Freezes

My Gentoo system is experiencing frequent and complete system freezes. When a freeze occurs, the entire computer becomes unresponsive – the screen freezes, and I cannot use the keyboard or mouse. The only way to recover is to power off the computer and restart it.

When do the freezes happen? The freezes seem to happen somewhat randomly during normal desktop use. I have not been able to reproduce the freezes with a dedicated GPU stress test.

What happens during a freeze?

• The screen image freezes completely.
  
• The computer becomes completely unresponsive to any input. Keyboard presses and mouse movements have no effect.
  
• The only way to regain control is to perform a hard power cycle.

How often do the freezes occur? The freezes are happening multiple times per day, which makes my system unreliable for daily use.

Troubleshooting Steps I Have Already Tried:

So far, I have tried the following kernel parameters:

amdgpu.aspm=0 (Disable PCIe Active State Power Management):
Result: Using amdgpu.aspm=0 might have slightly reduced how often the freezes happen, but the problem is still present.

amdgpu.dpm=0 (Disable AMDGPU Dynamic Power Management):
Result: System will not boot. When I add amdgpu.dpm=0 to my kernel parameters, my system fails to boot properly. It stops booting early in the process, and I have to remove the parameter to boot again.

radeon.dpm=1 and amdgpu.aspm=0 (Use Radeon DPM, disable AMDGPU ASPM):
Result: Freezes still happen. Trying to use the radeon driver's dynamic power management with radeon.dpm=1 did not fix the freezing problem.

Tests I Have Performed:

• Memory Test (Memtest86+): I have run Memtest86+ to check my system's RAM. The test completed with no errors.
  
• GPU Stress Test: I performed a GPU stress test with stress-ng to see if I could trigger a freeze under heavy GPU load. The freeze is not reproducible with the stress test.
  
• Temperature Monitoring: I have checked my CPU and GPU temperatures using system sensors. My system is not overheating before or during the freezes.

Relevant Log Information:

Here are some messages I have noticed in my system logs that might be related:

Boot Log Errors (These messages appear every time I boot, even with kernel parameters):

[Anon-E-moose]

It looks like you have secure display set, I would turn it off as I don't believe it necessary.

DRM_AMD_SECURE_DISPLAY is probably set to Y

Try it with that disabled (rebuild kernel and modules) and let us know what happened.
If it still doesn't work then wgetpaste your .config file
_________________
UM780, 6.12 zen kernel, gcc 13, openrc, wayland

[zedxot]

[Anon-E-moose]

You have a lot of stuff checked that I'm not sure you need, Selinux options, IMA options, etc.
Multiple displays selected amdgpu, radeon, looks like maybe i915, etc.

What hardware do you have? Model of laptop or motherboard model and any video cards that have been added (not on-board)
_________________
UM780, 6.12 zen kernel, gcc 13, openrc, wayland

[zedxot]

[Anon-E-moose]

Lets try this, there should be a file /etc/modprobe.d/blacklist.conf
edit that and add blacklists for radeon, radeonsi, i915 and reboot, does that change things?
_________________
UM780, 6.12 zen kernel, gcc 13, openrc, wayland

[zedxot]

[Anon-E-moose]

Do you have any of these 4?

[zedxot]

[tomtom69]

I might have the same issue with an AMD 3350G.
Can you check whether your system is accessible via ssh?
When the error occurs here I am able to connect via ssh (even if it is very slow) and I can see a process with kworker/u32:0+amdgpu-reset-dev consuming 100% CPU.
I downgraded kernel to 6.6.74 and upgraded mesa to 25.0.1 due to this bug report:
https://bugs.gentoo.org/947181
At the moment I do not have any freezes, but I did not test for a longer time.
Maybe downgrading kernel and upgrading mesa is worth a try.

[zedxot]

[Anon-E-moose]

You may need to blacklist other modules. Keep an eye on dmesg output for things not working right.
_________________
UM780, 6.12 zen kernel, gcc 13, openrc, wayland

[pietinger]