Re: Using __builtin_expect (likely/unlikely macros)

[Stefan Monnier]

>>> These macro calls would not help near calls to emacs_abort, as it
>>> should already be obvious to a careful human reader that the jump to
>>> emacs_abort is the road less traveled. (That's also obvious to GCC,
>>> since emacs_abort is _Noreturn.)
>> To human readers, yes, but from what I can tell, GCC is mixed on this.
> Then we should fix GCC, if the code it generates has a performance problem
> (whatever it is, it's quite small).

FWIW, the fact that a function is "_Noreturn" doesn't necessarily mean
that a call to it is unlikely (in many cases it is, I guess, but
definitely not all), so maybe GCC maintainers consciously decided not to
link the two.

> The GCC manual recommends against manually inserting such calls;

It's likely based on some past experiments that showed programmers
aren't very good at understanding what is likely and what isn't in
their code.

BTW I think instead of marking branches as likely or unlikely, I'd
prefer to tell GCC that some functions "should be slow"
(e.g. emacs_abort) so it optimizes the code paths that don't go through
those functions to the detriment of those that do.


        Stefan