SegFault libnvidia-glcore.so.560.35.03 - Arch 560.35.03 GTX3080

Issue occurs when starting Bottles in a wayland session on the 560.35.03 driver.

Example of CLI Error (similar happens when running vulkaninfo however it does not trigger a segfault):

(bottles:12111): Gdk-WARNING **: 16:44:40.551: vkCreateWaylandSurfaceKHR(): A host memory allocation has faile
d. (VK_ERROR_OUT_OF_HOST_MEMORY) (-1)
fish: Job 1, ‘bottles’ terminated by signal SIGSEGV (Address boundary error)

Examples of DMESG output below:

[ 292.538806] traps: bottles[5807] general protection fault ip:7d92b1741ed0 sp:7ffc124c1b48 error:0 in libnvidia-glcore.so.560.35.03[d41ed0,7d92b0a00000+1eab000]

bottles[8003]: segfault at 10 ip 00007a9a8e141ed0 sp 00007ffdb9123dc8 error 4 in libnvidia-glcore.so.560.35.03[d41ed0,7a9a8d400000+1eab000] likely on CPU 7 (core 1, socket 0)

SystemInfo:

OS: Garuda Linux x86_64 (Arch)
Host: XPS 15 9570
Kernel: 6.10.10-zen1-1-zen
Resolution: 1920x1080 5120x1440
DE: KDE Plasma 6.1.5
WM: KWin (Wayland)
CPU: Intel(R) Core(TM) i7-8750H (12) @ 4.10 GHz
GPU: NVIDIA GeForce GTX 1050 Ti Mobile [Discrete]
GPU: NVIDIA GeForce RTX 3080 Lite Hash Rate [Discrete]
GPU: Intel UHD Graphics 630 @ 1.10 GHz [Integrated]
Memory: 6553 MiB / 39901 MiB

nvidia-bug-report.log.gz (1.3 MB)

I have a similar issue when resuming from sleep; ever since the update to the latest driver (560.35.03) on Arch Linux, resume from sleep takes a long time. Sometimes it comes up, other times I have to switch to a console and restart lightdm and/or Cinnamon. Dmesg has this nugget during the resume:

kernel: cinnamon[998]: segfault at 0 ip 00007ea5fe941479 sp 00007ffd26840490 error 6 in libnvidia-glcore.so.560.35.03[b41479,7ea5fe200000+c00000] likely on CPU 4 (core 2, socket 0)

This is a Legion 5 Pro with 5800h and RTX3070 MaxQ, running in discrete graphics mode, and Cinnamon on X11 (not Wayland).