radv: Cyberpunk 2077 2.0 crashes X session
Cyberpunk 2077 2.0 frequently crashes (within five minutes usually) with no discernible pattern. RT is activated. Can get through the title screen, but usually crashes as soon as actual gameplay starts or shortly after. Seems to happen regardless of Proton version. Before the 2.0 update it was working reasonably well. Upon crashing the audio keeps playing for a while and after that I get sent back to GDM, whereas with Wayland I have to force reboot.
System information
- OS: Manjaro Linux
- GPU: NAVI 21 [Radeon RX 6950 XT]
- Kernel version: 6.5.5-1-MANJARO (happens also with 6.4 and 6.1)
- Mesa version: 23.1.8-manjaro1.1
- Xserver version (if applicable): 1.21.1.8
- Desktop manager and compositor: Gnome 44.5, mutter
- Wine/Proton version: Happens with Proton experimental, Proton 8.0, Proton 7.0, GE-Proton 7.35
- Environment variables: RADV_PERFTEST=rt, VKD3D_CONFIG=dxr11, DXVK_ASYNC=1, VKD3D_FEATURE_LEVEL=12_2
- Launch flags:-dx12 is the only relevant one, but does not seem to make a difference.
Regression
Used to work with Cyberpunk pre-2.0 update, regardless of Mesa version
Log files as attachment
dmesg:
[ 1900.944739] amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:1 pasid:32792, for process GameThread pid 7022 thread GameThread pid 7022)
[ 1900.944747] amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x00008000aa16f000 from client 0x1b (UTCL2)
[ 1900.944750] amdgpu 0000:03:00.0: amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x00101431
[ 1900.944752] amdgpu 0000:03:00.0: amdgpu: Faulty UTCL2 client ID: SQC (data) (0xa)
[ 1900.944754] amdgpu 0000:03:00.0: amdgpu: MORE_FAULTS: 0x1
[ 1900.944755] amdgpu 0000:03:00.0: amdgpu: WALKER_ERROR: 0x0
[ 1900.944756] amdgpu 0000:03:00.0: amdgpu: PERMISSION_FAULTS: 0x3
[ 1900.944758] amdgpu 0000:03:00.0: amdgpu: MAPPING_ERROR: 0x0
[ 1900.944759] amdgpu 0000:03:00.0: amdgpu: RW: 0x0
[ 1900.944764] amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:1 pasid:32792, for process GameThread pid 7022 thread GameThread pid 7022)
[ 1900.944767] amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x00008000e3d07000 from client 0x1b (UTCL2)
[ 1900.944769] amdgpu 0000:03:00.0: amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x00000000
[ 1900.944770] amdgpu 0000:03:00.0: amdgpu: Faulty UTCL2 client ID: CB/DB (0x0)
[ 1900.944771] amdgpu 0000:03:00.0: amdgpu: MORE_FAULTS: 0x0
[ 1900.944773] amdgpu 0000:03:00.0: amdgpu: WALKER_ERROR: 0x0
[ 1900.944774] amdgpu 0000:03:00.0: amdgpu: PERMISSION_FAULTS: 0x0
[ 1900.944775] amdgpu 0000:03:00.0: amdgpu: MAPPING_ERROR: 0x0
[ 1900.944777] amdgpu 0000:03:00.0: amdgpu: RW: 0x0
[ 1900.944781] amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:1 pasid:32792, for process GameThread pid 7022 thread GameThread pid 7022)
[ 1900.944784] amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x00008000aa16f000 from client 0x1b (UTCL2)
[ 1900.944786] amdgpu 0000:03:00.0: amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x00000000
[ 1900.944787] amdgpu 0000:03:00.0: amdgpu: Faulty UTCL2 client ID: CB/DB (0x0)
[ 1900.944789] amdgpu 0000:03:00.0: amdgpu: MORE_FAULTS: 0x0
[ 1900.944790] amdgpu 0000:03:00.0: amdgpu: WALKER_ERROR: 0x0
[ 1900.944791] amdgpu 0000:03:00.0: amdgpu: PERMISSION_FAULTS: 0x0
[ 1900.944792] amdgpu 0000:03:00.0: amdgpu: MAPPING_ERROR: 0x0
[ 1900.944792] amdgpu 0000:03:00.0: amdgpu: RW: 0x0
[ 1962.281561] [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring comp_1.2.0 timeout, signaled seq=10444, emitted seq=10445
[ 1962.281764] [drm:amdgpu_job_timedout [amdgpu]] *ERROR* Process information: process GameThread pid 7022 thread GameThread pid 7022
After this the GPU resets according to dmesg.
Edited by Francesco Fiusco