Yeah, so.
Execution model: process kernel (one stack per thread) <---> interrupt kernel (one stack per CPU, continuations)
These are quite deeply discussed in e.g. atomic-osdi99.pdf excerpts from which you can find
here.
You could probably trace related literature from that document, or i can dig more from my collection if you're interested.
Quote:
Some processor architectures inherently favor the process model and process model kernels are easier to make fully preemptible. Although full preemptibility comes at a cost, this cost is associated with preemptibility, not with the process model itself. Process model kernels use more per-thread kernel memory, but this is unlikely to be a problem in practice except for power-constrained systems
I'm not sure where the "event" model comes in your text, I'd need more context or the reference to the actual paper to be able to help.