Submit Feedback

SE2 - Linux/Proton AMDGPU Crash: [gfxhub] page fault

CaptainPicarl shared this bug ago

Submitted

While playing the new survival campaign on Proton Hotfix, I've been experiencing some pretty consistent amdgpu crashes. Journal shows:

Dec 02 20:18:57 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:2 pasid:32789)
Dec 02 20:18:57 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu:  Process Main Thread pid 44206 thread vkd3d_queue pid 44441
Dec 02 20:18:57 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu:   in page starting at address 0x00008006034d0000 from client 10
Dec 02 20:18:57 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x00201030
Dec 02 20:18:57 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu:          Faulty UTCL2 client ID: TCP (0x8)
Dec 02 20:18:57 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu:          MORE_FAULTS: 0x0
Dec 02 20:18:57 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu:          WALKER_ERROR: 0x0
Dec 02 20:18:57 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu:          PERMISSION_FAULTS: 0x3
Dec 02 20:18:57 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu:          MAPPING_ERROR: 0x0
Dec 02 20:18:57 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu:          RW: 0x0
Dec 02 20:19:07 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu: Dumping IP State
Dec 02 20:19:07 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu: Dumping IP State Completed
Dec 02 20:19:07 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu: [drm] AMDGPU device coredump file has been created
Dec 02 20:19:07 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu: [drm] Check your /sys/class/drm/card1/device/devcoredump/data
Dec 02 20:19:07 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu: ring gfx_0.0.0 timeout, signaled seq=4114113, emitted seq=4114115
Dec 02 20:19:07 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu:  Process Main Thread pid 44206 thread vkd3d_queue pid 44441
Dec 02 20:19:07 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu: Starting gfx_0.0.0 ring reset
Dec 02 20:19:07 {SystemName} kernel: amdgpu 0000:0b:00.0: amdgpu: Ring gfx_0.0.0 reset succeeded

Unfortunately, that devcoredump/data file didn't exist when I went to go find it. I did, however, have proton logging enabled. It's decently large - so let me know if you're interested in it at all.

Can't recall whether or not KSW is interested in supporting Proton / Linux, but I figure you can set me straight on that!

Love the new SE2. Hope to play it lots once it gets stable!

I have the same bug

Replies (1)

Spying Wind ●

I have an AMD GPU and I haven't experienced this since using this guide from SE1: Setting up Space Engineers in Linux (in 2025)

All I did was:

Extract the tar.gz from GE-Proton10-27 into ~/.steam/root/compatibilitytools.d/
Force compatibility with GE-Proton10-27
Restart Steam.

The Launch Options I use:

DOTNET_ROOT="" DXVK_NVAPI_ALLOW_OTHER_DRIVERS=1 PROTON_ENABLE_NVAPI=1 PROTON_ENABLE_WAYLAND=1 gamemoderun %command%

The game still crashes, but it hasn't taken down my desktop environment as of yet.

Hope that help.

I have an AMD GPU and I haven't experienced this since using this guide from SE1: Setting up Space Engineers in Linux (in 2025)

All I did was:

Extract the tar.gz from GE-Proton10-27 into ~/.steam/root/compatibilitytools.d/
Force compatibility with GE-Proton10-27
Restart Steam.

The Launch Options I use:

DOTNET_ROOT="" DXVK_NVAPI_ALLOW_OTHER_DRIVERS=1 PROTON_ENABLE_NVAPI=1 PROTON_ENABLE_WAYLAND=1 gamemoderun %command%

The game still crashes, but it hasn't taken down my desktop environment as of yet.

Hope that help.

Reply URL

Spying Wind ●

There where some crashes. I've found that adding 'RADV_DEBUG=hang,allbos' seems to have fixed all crashing and gpu resetting issues. With just 'hang' it prevented KDE from crashing. With 'allbos' it stopped all reset related crashes all together.

What does the `allbos` option do? "force all allocated buffers to be referenced in submissions" - mesa3d

URL