Arch Linux

ghostboy.sh

Member

Registered: 2025-07-29

Posts: 1

[amdgpu] Fatal error during GPU init - hang on boot

Hi all,
I've been using arch as my daily driver since May and it has worked great. A couple of weeks ago I found that my machine wouldn't wake after automatically being put into suspend mode. When booting, it would hang on `triggering uevents...`. I used Windows for a few days and then suddenly Arch began booting successfully again. Now, the issue has returned for the last week without any system config changes and I'm seemingly unable to boot from any Linux distro (arch, arch-lts, arch ISO USB, Ubuntu).

I’m consistently getting a fatal error during GPU init, and boot hangs on this line when using `loglevel=7`:

amdgpu: Fetched VBIOS from ROM BAR
amdgpu: ATOM BIOS: ATOMBIOSBK-AMD VERO17.001.000.049.000000

I'm only able to get arch to boot if I use `nomodeset` and I verified linux-firmware, amd-ucode, mesa-utils, and mesa are all installed and up to date + rebuilt initramfs with `mkinitcpio -P`. Booting with `amdgpu.dc=0` still hangs.

Here are my hardware details:

OS: Arch Linux x86_64
Host: MS-7B79 (2.0)
Kernel: Linux 6.12.40-1-lts
Uptime: 20 mins
Packages: 856 (pacman)
Shell: bash 5.3.3
Display (Unknown-1): 1024x768 @ 60 Hz in 13"
WM: Hyprland 0.50.1 (Wayland)
Theme: Adwaita [GTK2/3/4]
Icons: Adwaita [GTK2/3/4]
Font: Adwaita Sans (11pt) [GTK2/3/4]
Cursor: default (24px)
Terminal: kitty 0.42.2
Terminal Font: JetBrainsMonoNF-Regular (10pt)
CPU: AMD Ryzen 7 2700 (16) @ 3.20 GHz
GPU: AMD Radeon RX 5700 XT
Memory: 3.56 GiB / 31.29 GiB (11%)
Swap: Disabled
Disk (/): 32.83 GiB / 456.89 GiB (7%) - ext4

This is what I see in `journalctl -b -1 -p err`:

Jul 28 17:07:19 arch-ssd kernel: hid-generic 0003:1532:024E.0004: No inputs registered, leaving
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: SMU: response:0xFFFFFFFF for index:6 param:0x00000000 message:EnableAllSmuFeatures?
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: Failed to enable requested dpm features!
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: Failed to setup smc hw!
Jul 28 17:07:19 arch-ssd kernel: [drm:amdgpu_device_init.cold [amdgpu]] *ERROR* hw_init of IP block <smu> failed -121
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: amdgpu_device_ip_init failed
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: Fatal error during GPU init
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: ring_buffer_start = 00000000dae4874a; ring_buffer_end = 000000005673fb1c; write_frame = 00000000ea1dd0bb
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: write_frame is pointing to address out of bounds
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: ring_buffer_start = 00000000dae4874a; ring_buffer_end = 000000005673fb1c; write_frame = 00000000ea1dd0bb
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: write_frame is pointing to address out of bounds
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: ring_buffer_start = 00000000dae4874a; ring_buffer_end = 000000005673fb1c; write_frame = 00000000ea1dd0bb
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: write_frame is pointing to address out of bounds
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: ring_buffer_start = 00000000dae4874a; ring_buffer_end = 000000005673fb1c; write_frame = 00000000ea1dd0bb
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: amdgpu: write_frame is pointing to address out of bounds
Jul 28 17:07:19 arch-ssd kernel: amdgpu 0000:29:00.0: probe with driver amdgpu failed with error -121

Any insight would be so very appreciated. Thank you!

jaller94

Member

Registered: 2025-07-30

Posts: 1

Re: [amdgpu] Fatal error during GPU init - hang on boot

Hi,
I'm struggling with what I think might be the same issue. It's a Lenovo ThinkPad P14s with an AMD Ryzen Pro 7.
Since the first week of July, I'm only able to boot with `nomodeset`, suspense freezes the laptop and CPU usage is unusually high when any kind of video decoding is taking place.

Another thread I have my eyes on is https://bbs.archlinux.org/viewtopic.php?id=306823 where the issue seems to have been resolved "after reinstalling the bootloader, regenerating the configs, initramfs, etc".

I have not tried this yet, but vimlucid said that they have been able to get past the partition encryption, after "Triggering uevents…", when the screen turns black, by blindly typing in their password. "voila - I'm in!"

Last edited by jaller94 (2025-07-30 16:32:49)