Async and scheduling - how do libraries avoid blocking at the lowest level?

Question

I\'ve been using various concurrency constructs for a while now without much consideration for how all the magic happens, which has recently made me increasingly uneasy.