GPU randomly unloading + dmesg spammed with errors since 535.98 update

Kind of errors I’m getting:

NVRM rmapiAllocWithSecInfo: RMAPI_GPU_LOCK_INTERNAL alloc requested without holding the RMAPI lock
NVRM serverFreeResourceTree: hObject 0x0 not found for client 0xc1d00080

The GPU gets randomly unloaded from my system, which is very awkward (this has never happened since I’m using nvidia+linux)

nvidia-bug-report.log.gz (217.1 KB)

Just faced this unloading thing again, and the exact same messages in dmesg.
nvidia-bug-report.log.gz (229.1 KB)