Farming Simulator 25: crash when loading 4x maps

vkd3dopen

HansKristian-Work/vkd3d-proton#3103 · opened 2026-06-11 by KilianHanich · updated 2026-08-22 · 6 comments · github

KKilianHanich 2026-06-11 github

When trying to load into a 4x map (as in four times larger than the base game maps; I tested multiple), when the load gets to the "compiling shaders" step at the end, the game hands in dmesg shows the following output:

[21759.821359] amdgpu 0000:07:00.0: Dumping IP State
[21759.826946] amdgpu 0000:07:00.0: Dumping IP State Completed
[21759.827060] amdgpu 0000:07:00.0: [drm] AMDGPU device coredump file has been created
[21759.827064] amdgpu 0000:07:00.0: [drm] Check your /sys/class/drm/card0/device/devcoredump/data
[21759.827068] amdgpu 0000:07:00.0: ring gfx_0.0.0 timeout, signaled seq=85937, emitted seq=85939
[21759.827072] amdgpu 0000:07:00.0:  Process FarmingSimulato pid 101911 thread vkd3d_queue pid 102066
[21759.827075] amdgpu 0000:07:00.0: Starting gfx_0.0.0 ring reset
[21759.829297] amdgpu 0000:07:00.0: Ring gfx_0.0.0 reset succeeded
[21759.829300] amdgpu 0000:07:00.0: [drm] device wedged, but recovered through reset

It works fine while trying to load normal sized maps.

System information

GPU: AMD Radeon RX 6750XT
Driver: radv, Mesa 26.0.8
Wine version: wine-11.0 (Staging)
VKD3D-Proton version: 3.0.1

Log files

Attached is the log of VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation VKD3D_CONFIG=vk_debug VKD3D_DEBUG=warn wine FarmingSimulator2025.exe -skipStartVideos >game.log 2>&1.
I can also provide the device coredump of this run if wanted, but since I don't know if there's sensitive information in there, I will only do it on request.
All files/logs are from the same run.

wine.log

KKilianHanich 2026-06-11 github

additional info: Using Proton Experimental instead of Wine doesn't change anything about this.

Rrunar-work 2026-06-11 github

I can also reproduce this on RX 7600 with 7.0.11-arch1-1. I don't get this hang with AMDVLK. It doesn't seem to be a regression (Mesa main and 25.0.0, Proton 10 and Experimental bleeding-edge gives the same result), but the info I can gather seems a bit random so far.

It still hangs with RADV_DEBUG=hang, but all I got with 26.1.2 was a dump folder with an empty trace.log file. One of the hangs with breadcrumbs enabled just ended up spewing breadcrumbs into the log until it grew to over 1 GiB and I stopped the game. Manual dump with Mesa main made umr segfault, so I switched to umr-git and 25.0.0 since I've had more luck with older Mesa versions before. After the initial hang with GPU recovery disabled, the game kept running and I got a few more hangs with different wave dumps from each invocation until it settled. I also managed to relaunch the game after the initial hang, so that was also weird. The breadcrumbs from both launches point to different compute shaders, so I included both in case there are any clues there.

farming_simulator_25_manual_umr_dump_rx7600.zip

Rrunar-work 2026-06-12 github

I tried with the LTS kernel (6.18.35-1) to see if it would make a difference, and with GPU recovery disabled and RADV_DEBUG=hang RADV detects a hang in the Proton log and makes a dump folder with an empty trace.log file with both main and 25.0.0, but there are no amdgpu errors in the dmesg log and the game just freezes and can be stopped without issue via the Steam client. Without RADV_DEBUG=hang on LTS it gets past the shader compilation on the loading screen without a hang or freeze.

With validation layers enabled, there's no hang on 7.0 either, the game just freezes during shader compilation. GPU recovery disabling doesn't seem to work reliably at all, so I haven't been able to get two breadcrumbs traces that match, they seem to be random. Sometimes there's a hang in the dmesg log and I can just wait a little while and the GPU seems to recover, even though there's no reset.

[  708.542889] amdgpu 0000:03:00.0: Dumping IP State
[  708.543791] amdgpu 0000:03:00.0: Dumping IP State Completed
[  708.543796] amdgpu 0000:03:00.0: [drm] AMDGPU device coredump file has been created
[  708.543797] amdgpu 0000:03:00.0: [drm] Check your /sys/class/drm/card1/device/devcoredump/data
[  708.543799] amdgpu 0000:03:00.0: ring gfx_0.0.0 timeout, signaled seq=95862, emitted seq=95864
[  708.543801] amdgpu 0000:03:00.0:  Process FarmingSimulato pid 11651 thread vkd3d_queue pid 11727
[  708.543803] amdgpu 0000:03:00.0: GPU recovery disabled.
[  795.067206] input: Microsoft X-Box 360 pad 0 as /devices/virtual/input/input32
[  796.660058] amdgpu 0000:03:00.0: [drm] PCIE GART of 512M enabled (table at 0x00000081FEB00000).
[  796.660094] amdgpu 0000:03:00.0: PSP is resuming...
[  796.708726] amdgpu 0000:03:00.0: reserve 0x1300000 from 0x81fc000000 for PSP TMR
[  796.803415] amdgpu 0000:03:00.0: RAS: optional ras ta ucode is not available
[  796.810828] amdgpu 0000:03:00.0: RAP: optional rap ta ucode is not available
[  796.810830] amdgpu 0000:03:00.0: SECUREDISPLAY: optional securedisplay ta ucode is not available
[  796.810832] amdgpu 0000:03:00.0: SMU is resuming...
[  796.810835] amdgpu 0000:03:00.0: smu driver if version = 0x00000035, smu fw if version = 0x00000040, smu fw program = 0, smu fw version = 0x00525f00 (82.95.0)
[  796.810837] amdgpu 0000:03:00.0: SMU driver if version not matched
[  796.841361] amdgpu 0000:03:00.0: SMU is resumed successfully!

Rrunar-work 2026-06-15 github

Here's the full log I got from one of the hangs with the breadcrumbs spamming and one with breadcrumbs_trace for comparison:

steam-2300320_breadcrumbs_spam.zip

steam-2300320_breadcrumbs_trace_spam.tar.gz

Rrunar-work 2026-08-14 github

This should be fixed now with https://github.com/HansKristian-Work/vkd3d-proton/pull/3203, i.e. on Proton Experimental bleeding-edge.

KKilianHanich 2026-08-22 github

So, I compiled and installed vkd3d-proton manually, and after testing it the same error still occurred.