Arch Linux

Stef9998

Member

From: de

Registered: 2019-01-13

Posts: 6

Random Freezes (Black Screen)

Hi, I got the problem, that my system randomly gets a black screen and everything I can do is to forcefully shut down by using the power button.

The problem started in June, but I had stuff to do, so I used my laptop in that time for work and didn't care about the problem for some time.

Yesterday I tried to use my Arch PC again and it seemed to work fine after I updated everything.
But after some hours running the system black-screened again.

Here is the output of journalctl from yesterday where I couldn't find something that jumped out:
journalctl_230905

But I gotta say I have no idea where to even start to look for narrowing down what the problem is.

Thanks in advance :)

Edit: froze again an hour ago (around 14:25). I let the PC run for some more time and force shut down it around 15 o'clock.
journalctl_230906

It seems to be some problem with the GPU.

The newest two outputs from journalctl from june seem to have a simular error:
journalctl_230620_1
journalctl_230620_2

Last edited by Stef9998 (2023-09-06 13:25:46)

---

Arch Linux, it's not perfect, but it's where I stopped distro-hopping.

resit4

Member

Registered: 2023-09-07

Posts: 1

Re: Random Freezes (Black Screen)

The blackscreen starts with amdgpu being unable to reset the GPU:

Sep 06 14:22:55 stefan-desk-arch kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx timeout, signaled seq=1033910, emitted seq=1033912
Sep 06 14:22:55 stefan-desk-arch kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* Process information: process picom pid 1606 thread picom:cs0 pid 1645
Sep 06 14:22:55 stefan-desk-arch kernel: amdgpu 0000:01:00.0: amdgpu: GPU reset begin!
Sep 06 14:22:55 stefan-desk-arch kernel: amdgpu 0000:01:00.0: amdgpu:
                                         last message was failed ret is 65535

Someone worked around a similar issue by lowering the clock speed here
It was apparently caused by the GPU overheating.

This might not be related, but still worth trying:

In this forum post,
someone was able to let the GPU recover by adding the kernel parameters:

amdgpu.gpu_recovery=1 amdgpu.lockup_timeout=3000

and installing the AUR package amdgpu-fan.