Fixed.
Somewhere in upgrading nvidia drivers to the 520 branch, kernel to 5.19, wine to lutris-GE-Proton7-32-x86_64, and vkd3d-proton to 2.7, this location / save point no longer crashes.
Happy for this issue to be closed.
On a side note, VKD3D_CONFIG=dxr11 now works in this game, where previously only dxr worked
ge-proton7-32x1 2022-10ge-proton7-26x1 2022-09proton 2.6x1 2022-09VKD3D_CONFIG=dxr11x1 2022-10
Mortal Shell crash with "0430:err:vkd3d_wait_for_gpu_timeline_semaphore: Failed to wait for Vulkan timeline semaphore, vr -4." multiple entries per crash in the log.
Only happens for me when all of the following criteria are met:
Crash occurs during a normal black screen between the main menu and the actual game.
Hours of play through in other areas have seen zero crashes, and checking the logs, zero other instances of "Failed to wait for Vulkan timeline semaphore" except this scenario.
vr -4 appears to be vkResult = VK_ERROR_DEVICE_LOST
Once the game has crashed, the OS and everything else is fine, suggesting the physical device is ok, and it's just as logical device lost.
However, kernel output is: "NVRM: Xid (PCI:0000:21:00): 31, pid=7497, name=Dungeonhaven-Wi, Ch 00000022, intr 00000000. MMU Fault: ENGINE HOST3 HUBCLIENT_ESC faulted @ 0x0_3edbf000. Fault is of type FAULT_PTE ACCESS_TYPE_VIRT_WRITE"
Xid error 31 = "GPU memory page fault"
Hardware: AMD Ryzen 3970x, 128GB RAM, Nvidia RTX 3090
OS & Drivers: Linux Mint 20.3, Kernel 5.15, Nvidia 515.65.01
Wine: lutris-GE-Proton7-26-x86_64, vkd3d-proton 2.6, dxvk-nvapi 0.5.4, dxvk 1.10.3
Game: Mortal Shell - GOG version - Build 08.25.21 Revision 1.014707
Screenshot showing the location which triggers the crash if criteria above are met, taken with "Shadow Quality" at "Medium"
Any advice on how I can gain further relevant info with logging or tracing?