[SOLVED] System stuck after amdgpu initializes

AngelBePro · 2025-05-22 21:51:16

When I boot my PC I randomly get a black screen, so I decided to set loglevel to debug (8) and see what's going on. https://ebise.online/$/vXENj hope it's not a burden but I am sending what I got as a image since journalctl can't log this issue. Under windows, I get a blue screen with the error DRIVER VERIFIER DMA VIOLATION. Hope that is enough info.

Last edited by AngelBePro (2025-05-23 17:41:23)

seth · 2025-05-23 07:56:08

3rd link below. Mandatory.
Disable it (it's NOT the BIOS setting!) and reboot windows and linux twice for voodo reasons.

"amdgpu.runpm=0 amdgpu.aspm=0 amdgpu.bapm=0", https://wiki.archlinux.org/title/Kernel_parameters
Do you only boot the multi-user.target or can you switch the VT from a black screen in case you're booting straight into a display server?
How frequent is this?

AngelBePro · 2025-05-23 12:11:10

It's as I said kinda random, but I am not so sure how much windows is causing it, as the issue occurs without going into windows sometimes. But it may also be it as I disabled fast start and even tried the kernel parameters, and the system is still black screening. It doesnt let me switch the VT. One thing that might fix the issue is instead of using the reset button to reboot, I instead shut down, and then try again. For frequency, it's not that common, something like a 10% chance it feels like. Not sure if this helps, but my launch parameters are:

loglevel=3 quiet sysrq_always_enabled=1 resume=/dev/nvme0n1p2 drm.edid_firmware=DP-2:edid/edid.bin

Last edited by AngelBePro (2025-05-23 12:20:21)

AngelBePro · 2025-05-23 12:23:34

Ok I realized what the issue was with the log after looking at my launch parameters. I removed quiet mode and turns out, a start job is running for /dev/nvme0n1p2. Removing the resume parameter fixed it. I expected that hibernating will not work without resume, but it still works. Is there anything else I should do?

Last edited by AngelBePro (2025-05-23 12:28:14)

seth · 2025-05-23 13:25:23

https://wiki.archlinux.org/title/Power_ … ibernation
systemd should™ be able to resume from hibernation w/o explicitly listing the image location on EFI systems, but there's actually an active thread where this occasionally fails (so far not clear whether it's systemd's fault)
If you define the location of the hibernation image, make sure it's the correct one and DO NOT REFERENCE IT VIA THE DEVICE NODE, ie. don't use sth. like /dev/nvme0n1p2 but the UUID.

Please always remember to mark resolved threads by editing your initial posts subject - so others will know that there's no task left, but maybe a solution to find.
Thanks.

AngelBePro · 2025-05-23 17:41:08

Marking as solved until I experience this issue again, if that happens.

Arch Linux

#1 2025-05-22 21:51:16

[SOLVED] System stuck after amdgpu initializes

#2 2025-05-23 07:56:08

Re: [SOLVED] System stuck after amdgpu initializes

#3 2025-05-23 12:11:10

Re: [SOLVED] System stuck after amdgpu initializes

#4 2025-05-23 12:23:34

Re: [SOLVED] System stuck after amdgpu initializes

#5 2025-05-23 13:25:23

Re: [SOLVED] System stuck after amdgpu initializes

#6 2025-05-23 17:41:08

Re: [SOLVED] System stuck after amdgpu initializes

Board footer