On 02/02/2018 19:50, Linus Torvalds wrote: > On Fri, Feb 2, 2018 at 6:59 AM, David Woodhouse <dwmw@xxxxxxxxxxxx> wrote: >> With retpoline, tight loops of "call this function for every XXX" are >> very much pessimised by taking a prediction miss *every* time. >> >> This one showed up very high in our early testing, and it only has five >> things it'll ever call so make it take an 'op' enum instead of a >> function pointer and let's see how that works out... > Umm. May I suggest a different workaround? > > Honestly, if this is so performance-critical, the *real* fix is to > actually just mark all those "slot_handle_*()" functions as > "always_inline". I replied quickly from the phone before reading the rest of the thread---yeah, always_inline is the way to go. I see the same differences as Linus and David (slight improvement for slot_handle_*, +1k if you add kvm_handle_hva and kvm_handle_hva_range). At least for slot_handle_* it's a no-brainer. The others are basically the MMU notifier implementation; in the perfect case it should actually never be called (or at least it ought to be very rare), so I think we can keep the indirect calls for now. Paolo
