These are things that kernel developers are surely mindful of when coming up with and implementing eBPF functionality. Regardless, I'm sure I've run this same

These are things that kernel developers are surely mindful of when coming up with and implementing eBPF functionality. Regardless, I'm sure I've run this same test years ago and seen the system call count still in the same order (that is, a couple of million per second). I really doubt Spectre mitigations are what are causing what should be a few dereferences and function calls to take around a thousand clock cycles.