RADV: Hollow Knight: blocky screen
After playing for (usually) a few hours, the screen sometimes becomes blocky when going into a new area. When going back to the previous area, that area looks fine for a minute or so, but after that, that area becomes blocky as well. Going back to the main menu fixes the issue. I had encountered this about 5 times in my 22 hour playtime, but have now found an area where I can reproduce it fairly quickly. In this area, the game also crashes shortly after the screen becomes blocky most of the times.
Log files (for system lockups / game freezes / crashes)
- Relevant dmesg:
[49655.207975] [drm:amdgpu_cs_ioctl [amdgpu]] *ERROR* Failed to initialize parser -2!
- Game log: Player.log
Steps to reproduce
Savegame, belongs in
$HOME/.config/unity3d/Team Cherry/Hollow Knight user1.dat
Video on how to get to the area (it didn't become blocky instantly when taking the video, but after I stopped recording and played for another minute it became blocky) blocky_area.mkv
System: Host: exolyte Kernel: 5.15.11-arch2-1 x86_64 bits: 64 compiler: gcc v: 11.1.0 Desktop: MATE 1.26.0 wm: marco dm: LightDM Distro: Antergos Linux base: Arch Linux CPU: Info: quad core model: Intel Core i7-2600 bits: 64 type: MT MCP arch: Sandy Bridge rev: 7 cache: L1: 256 KiB L2: 1024 KiB L3: 8 MiB Speed (MHz): avg: 3663 high: 3703 min/max: 1600/3800 cores: 1: 3608 2: 3686 3: 3703 4: 3675 5: 3680 6: 3646 7: 3670 8: 3642 bogomips: 54301 Flags: avx ht lm nx pae sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx Graphics: Device-1: AMD Ellesmere [Radeon RX 470/480/570/570X/580/580X/590] vendor: ASUSTeK driver: amdgpu v: kernel bus-ID: 01:00.0 chip-ID: 1002:67df Display: x11 server: X.Org 18.104.22.168 compositor: marco driver: loaded: amdgpu resolution: 1920x1080~60Hz s-dpi: 96 OpenGL: renderer: AMD Radeon RX 570 Series (POLARIS10 DRM 3.42.0 5.15.11-arch2-1 LLVM 13.0.0) v: 4.6 Mesa 21.3.2 direct render: Yes
I have 4GB VRAM.
API captures (if applicable, optional)
Renderdoc (taken with renderdoc-bin v1.17-1 from AUR): https://mega.nz/file/iR8mmQqR#H3yi_TJG6N-fxRhw9kQNem93o9etfF6DFcri9wAINeE
Further information (optional)
Does the issue reproduce with
- the LLVM backend (
- or on the AMDGPU-PRO drivers?: yes
It also happens on
Some other people with the same issue on other hardware: