It turns out (once again) my scheduler issue is just timing not a “real” deadlock. So changing the frequency at which the scheduler is invoked is a temporary workaround but it looks like I just need to make some general improvements to scheduler performance in the longer term.
It was also a good opportunity to test my backup scheduler implementation, so it looks like it will be very easy to switch between the simpler scheduler version (“round-robin”) & more full-featured versions (“priority-based” etc.) after a little more cleanup.
Latest screenshot running on top of the kernel:
Leave a Reply