[2.1.9] Sudden page faults in amdgpu driver on linux

gridstop · Post by **gridstop** » Wed Jul 01, 2026 12:35 am

Yes I know things would normally point to some kind of hardware/driver error, but:

I did a lot of testing on 2.1.7 without crashes, and reverting to 2.0.77 has no crashing, and even 2.1.8 has no crashing, but 2.1.9 crashes quite quickly both at the menu while updating/downloading mods, while sitting at main menu, and while playing. Obviously this isn't solely a factorio bug but it could be factorio accidentally triggering a bug in amdgpu that 2.0.77, 2.1.7, and 2.1.8 don't, for some reason?

Mainly posting this in case anybody else sees this on amdgpu/9070 series AMD cards with 2.1.9, perhaps we'll get some more points of reference.

There is no output related to the crash in factorio's log, I guess it dies before that can happen, but there's the main system info from the log:

Code: Select all

   0.000 2026-07-01 00:18:14; Factorio 2.1.9 (build 86829, linux64, steam, space-age)
   0.000 Operating system: Linux
   0.000 Initializing Steam API.
   0.000 Program arguments: "/home/user/.local/share/Steam/steamapps/common/Factorio/bin/x64/factorio" 
   0.000 Config path: /home/user/.factorio-2.1/config/config.ini
   0.000 Read data path: /home/user/.local/share/Steam/steamapps/common/Factorio/data
   0.000 Write data path: /home/user/.factorio-2.1 [564515/952843MB]
   0.000 Binaries path: /home/user/.local/share/Steam/steamapps/common/Factorio/bin
   0.006 System info: [CPU: AMD Ryzen 5 5600X 6-Core Processor, 12 cores, RAM: 48113 MB]
   0.006 Environment: DISPLAY=:1 WAYLAND_DISPLAY=wayland-0 DESKTOP_SESSION=plasma XDG_SESSION_DESKTOP=KDE XDG_CURRENT_DESKTOP=KDE SDL_VIDEODRIVER=<unset> __GL_FSAA_MODE=<unset> __GL_LOG_MAX_ANISO=<unset> __GL_SYNC_TO_VBLANK=<unset> __GL_SORT_FBCONFIGS=<unset> __GL_YIELD=<unset>
   0.027 Display options: [FullScreen: true] [VSync: true] [UIScale: automatic (100.0%)] [Native DPI: true] [Screen: 4294967295] [Special: lmw] [Lang: en]
   0.116 Video driver: wayland
   0.117 Available displays: 2
   0.117  [0]: Dell Inc. DELL G3223Q - {3, [0,0], 3200x1800, SDL_PIXELFORMAT_XRGB8888, 120.00 Hz}
   0.117  [1]: HP Inc. HP X24ih - {4, [3200,200], 1920x1080, SDL_PIXELFORMAT_XRGB8888, 120.00 Hz}
   0.146 Initialised OpenGL:[3] AMD Radeon RX 9070 XT (radeonsi, gfx1201, ACO, DRM 3.64, 7.0.14-arch1-1); driver: 4.6 (Core Profile) Mesa 26.1.3-arch1.2
   0.146   [Size] Window: 3200x1800, Backbuffer: 3840x2160
   0.146   [Extensions] s3tc:yes; KHR_debug:yes; ARB_clear_texture:yes, ARB_copy_image:yes, ARB_pipeline_statistics_query:yes, ARB_gl_spirv:yes, ARB_ES2_compatibility:yes
   0.146   [Version] 4.6
   0.162 Graphics settings preset: very-high
   0.163   Dedicated video memory size 16384 MB (detected from AMD Radeon RX 9070 XT (RADV GFX1201); VendorID: 0x1002; DeviceID: 0x7550)
   0.172 Graphics options: [Graphics quality: high] [Video memory usage: all] [DXT: high-quality]
   0.172                   [Max threads (load/render): 32/8] [Max texture size: 0] [Tex.Stream.: false] [Rotation quality: normal] [Other: STDCWTl] [B:0,C:0,S:100]
   0.186 [Audio] Driver:pulseaudio, Device:Starship/Matisse HD Audio Controller Digital Stereo (IEC958), Depth:32, Channels:2, Frequency:44100

Dmesg output:

Code: Select all

[41662.368485] amdgpu 0000:0b:00.0: [gfxhub] page fault (src_id:0 ring:24 vmid:6 pasid:2711)
[41662.368491] amdgpu 0000:0b:00.0:  Process factorio pid 69370 thread factorio:cs0 pid 72085
[41662.368494] amdgpu 0000:0b:00.0:   in page starting at address 0x0000000018000000 from client 10
[41662.368497] amdgpu 0000:0b:00.0: GCVM_L2_PROTECTION_FAULT_STATUS:0x00601431
[41662.368499] amdgpu 0000:0b:00.0:      Faulty UTCL2 client ID: SQC (data) (0xa)
[41662.368529] amdgpu 0000:0b:00.0:      MORE_FAULTS: 0x1
[41662.368530] amdgpu 0000:0b:00.0:      WALKER_ERROR: 0x0
[41662.368532] amdgpu 0000:0b:00.0:      PERMISSION_FAULTS: 0x3
[41662.368534] amdgpu 0000:0b:00.0:      MAPPING_ERROR: 0x0
[41662.368536] amdgpu 0000:0b:00.0:      RW: 0x0
[41662.368541] amdgpu 0000:0b:00.0: [gfxhub] page fault (src_id:0 ring:24 vmid:6 pasid:2711)
[41662.368543] amdgpu 0000:0b:00.0:  Process factorio pid 69370 thread factorio:cs0 pid 72085
[41662.368546] amdgpu 0000:0b:00.0:   in page starting at address 0x0000000018000000 from client 10
[41662.368551] amdgpu 0000:0b:00.0: [gfxhub] page fault (src_id:0 ring:24 vmid:6 pasid:2711)
[41662.368553] amdgpu 0000:0b:00.0:  Process factorio pid 69370 thread factorio:cs0 pid 72085
[41662.368555] amdgpu 0000:0b:00.0:   in page starting at address 0x0000000018000000 from client 10
[41662.368561] amdgpu 0000:0b:00.0: [gfxhub] page fault (src_id:0 ring:24 vmid:6 pasid:2711)
[41662.368563] amdgpu 0000:0b:00.0:  Process factorio pid 69370 thread factorio:cs0 pid 72085
[41662.368565] amdgpu 0000:0b:00.0:   in page starting at address 0x0000000018000000 from client 10
[41664.434838] amdgpu 0000:0b:00.0: Dumping IP State
[41664.436000] amdgpu 0000:0b:00.0: Dumping IP State Completed
[41664.436050] amdgpu 0000:0b:00.0: [drm] AMDGPU device coredump file has been created
[41664.436053] amdgpu 0000:0b:00.0: [drm] Check your /sys/class/drm/card1/device/devcoredump/data
[41664.436056] amdgpu 0000:0b:00.0: ring gfx_0.0.0 timeout, signaled seq=8219646, emitted seq=8219648
[41664.436061] amdgpu 0000:0b:00.0:  Process factorio pid 69370 thread factorio:cs0 pid 72085
[41664.436064] amdgpu 0000:0b:00.0: Starting gfx_0.0.0 ring reset
[41664.436185] amdgpu 0000:0b:00.0: Ring gfx_0.0.0 reset succeeded
[41664.436189] amdgpu 0000:0b:00.0: [drm] device wedged, but recovered through reset

EDIT: I should add, I did test and have crashing in 2.1.9, then reverted to 2.1.8 without rebooting or anything, and had no crashing. But since doing a clean reboot and going back to 2.1.9, I haven't had any more crashing today. So this could have been some totally random magic or obscure bug in amdgpu that I've just never seen before that happened to occur the day I upgraded to 2.1.9 but cleared with a reboot. Probably not worth spending any time on unless anybody else sees it.