https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86203
--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Martin Sebor from comment #4)
> You're right, a strict reading of the standard does imply that the pointer
> argument to strlen could point to the integer. I keep forgetting about this
> unfortunate loophole. Ironically, the equivalent code using snprintf could
> be optimized because the argument to %s "shall be a pointer to the initial
> element
> of an array of character type." I.e., it cannot point to an int object:
>
> int n;
>
> int f (char *s)
> {
> n = __builtin_snprintf (0, 0, "%s", s);
> return __builtin_snprintf (0, 0, "%s", s); // could be replaced by
> return n;
> }
I would guess it could be folded all the way to return 0?
