How to Fix [drm:amdgpu_job_timedout] and AMD GPU Hang Crashes in Linux

During demanding gaming sessions on Linux using Proton, Wine, or native Vulkan engines, owners of AMD Radeon graphics cards (RX 5000, 6000, 7000 series, or Steam Deck) frequently encounter sudden screen freezes followed by a crash to desktop. Inspecting system kernel messages via dmesg reveals the underlying hardware scheduler fault: [drm:amdgpu_job_timedout] ring gfx_0.0.0 timeout followed by [drm:amdgpu_device_gpu_recover] *ERROR* GPU reset failed.

This error occurs when a graphic command submitted to the GPU's execution ring fails to finish before the Linux kernel's Direct Rendering Manager (DRM) scheduler timeout expires. Common triggers include overly aggressive dynamic power management (DPM) states, race conditions in Mesa's RADV Vulkan compiler during complex ray tracing shaders, or PCIe bus voltage dips under sudden compute bursts.

🚀 Step-by-Step Fixes

Step 1: Verify the exact GPU hang signature in dmesg

Immediately following a game freeze or desktop crash, open a terminal window and filter the kernel buffer to confirm the AMDGPU scheduler failure:

# Filter recent dmesg logs for GPU timeouts
sudo dmesg -T | grep -E "amdgpu_job_timedout|GPU reset|ring gfx"

If the terminal displays messages matching:

[drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_0.0.0 timeout, signaled seq=...
[drm:amdgpu_device_gpu_recover [amdgpu]] *ERROR* GPU reset failed

Your issue is definitely confirmed as a driver/hardware scheduler timeout rather than an ordinary game script bug.

Step 2: Configure Kernel Parameters to stabilize AMDGPU DPM

On desktop distributions such as Arch Linux, Bazzite, Fedora, or Ubuntu, runtime power management can cause the GPU to throttle down voltages prematurely between frame rendering cycles:

1. Open your bootloader configuration file with root privileges:

   BASH Copy

   sudo nano /etc/default/grub
   

2. Locate GRUB_CMDLINE_LINUX_DEFAULT and append the following parameters inside the quotes:

   TEXT Copy

   amdgpu.dpm=1 amdgpu.runpm=0
   

   • amdgpu.dpm=1: Enforces deterministic dynamic power state transitions.
   • amdgpu.runpm=0: Disables runtime deep power sleep on dedicated desktop Radeon cards.

3. Regenerate the boot configuration for your distribution:

   BASH Copy

   # On Debian, Ubuntu, and Pop!_OS:
   sudo update-grub
   
   # On Arch Linux, EndeavourOS, and CachyOS:
   sudo grub-mkconfig -o /boot/grub/grub.cfg
   
   # On Fedora (UEFI systems):
   sudo grub2-mkconfig -o /etc/grub2-efi.cfg
   

Step 3: Set Steam launch options for Vulkan and VKD3D-Proton

While Mesa's ACO shader compiler is fast, certain complex shaders in DirectX 12 games can overwhelm the execution ring. You can configure safer fallback parameters on a per-game basis in Steam (Properties -> Launch Options):

RADV_DEBUG=noaccel,llvm %command%

Or for DirectX 12 titles encountering buffer synchronization hangs:

VKD3D_CONFIG=no_upload_hwwrite,dxr11 %command%

Step 4: Disable PCIe Active State Power Management (ASPM)

PCIe ASPM powers down the bus link to conserve energy when the graphics card enters brief idle periods. On certain motherboards, the latency required to wake the PCIe bus back to Full Speed exceeds the driver's timeout threshold:

Add the following kernel option to /etc/default/grub:

pcie_aspm=off

Regenerate your GRUB configuration and restart the PC. This maintains PCIe bus lanes at full operational readiness with zero power switching delay.

Frequently Asked Questions

Does this error mean my AMD graphics card is physically failing?

In over 95% of cases, this is strictly a software or firmware scheduling issue involving driver timeouts, kernel power management policies, or PCIe ASPM delays.

Does this problem happen on both Wayland and X11?

Yes, but recent Wayland protocols with Explicit Sync (introduced in Linux 6.8+ and Wayland Protocols 1.34) have significantly mitigated implicit synchronization hangs on Mesa drivers.