That's a pretty odd workaround considering that you can reserve cores to the point where not even kernel tasks run on them and then pin a single userspace thread to that core so it can run without ever being preemptively descheduled