Here are a few FAQs with regard to hardware interrupts and how they are handled. We haven't touched on these areas yet:
- On a multicore system, are all hardware interrupts routed to one CPU? If not, how are they load balanced? Can I change this?
- Does the kernel maintain a separate IRQ stack?
- How can I obtain metrics on interrupts? Can I measure interrupt latency?
The idea here is to provide brief answers; we encourage you to dig deeper and try things out for yourself! At the risk of repetition, remember, the empirical approach is best!