> Thanks for the explanations. I don’t think it would work (e.g., IRQs). I can > avoid generalizing and just detect the "magic sequence” of the code, but let > me give it some more thought.
If you ask me I would just use compiler profile feedback or autofdo (if your compiler has a working version) The compiler can do a much better job at optimizing this than you ever could. Manual FDO needs some kernel patching though. -Andi