[Bug c/71479] error on __builtin_add_overflow with bool or enum pointer as last argument

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71479

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I see that in __builtin_*_overflow_p I'm actually checking just INTEGRAL_TYPE_P
rather than INTEGER_TYPE for the last argument.
But, I think I'm in favor of tweaking the documentation and only allowing for
*_overflow pointer to T, for *_overflow_p T, where T is integral type other
than boolean or enumeration type.
While for _Bool/bool, it is not that hard to define what exactly it would mean
(check if the infinite precision result is 0 or 1 or not), for enums, would we
check if it fits into the underlying type, or fits into the strict precision of
the enum, have different behavior based on -fstrict-enums, or check if there is
any enumerator at that point, ... ?