Chrome on XWayland intermittent GPU process crash

Using Chrome on XWayland, the GPU process will occasionally crash. The log message from Chrome is [3322:3322:1028/215616.408344:ERROR:shared_context_state.cc(833)] : SharedContextState context lost via ARB/EXT_robustness. Reset status = GL_GUILTY_CONTEXT_RESET_KHR. This is accompanied by a kernel log message: kernel: NVRM: Xid (PCI:0000:01:00): 31, pid=3322, Ch 00000008, intr 10000000. MMU Fault: ENGINE GRAPHICS GPCCLIENT_PROP_0 faulted @ 0x1_148c0000. Fault is of type FAULT_PTE ACCESS_TYPE_WRITE. This issue does not occur on Xorg. Happens with Vulkan enabled or disabled. GPU compositing is enabled.

OS: Arch Linux
Driver version: 495.44
DE: GNOME 41

I would like to send the bug report log file privately, if possible, since it contains personal information.

You could email them linux-bugs AT nvidia.com

I think this might be related to a permissions issue on the video memory file.

I am able to mitigate this somewhat (but not completely) using the flag --use-angle=vulkan.

I was able to reproduce the issue on another computer.

Issue persists with driver 495.46

Issue persists with version 510.54

Its possible to reproduce this crash. Using Discord web version and clicking fullscreen on a video posted, the GPU process exits immediately. The video then plays after a black screen flicker during the restart of the process.

Issue occurs if repeatedly fullscreen is entered and exited.

Log contains [drm:nv_drm_gem_alloc_nvkms_memory_ioctl [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000100] Failed to allocate NVKMS memory for GEM object on every crash.