[Bug libfortran/125623] [17 Regression] write_float.def:44:37: error: implicit declaration o f function 'frexpl' [-Wimplicit-function-declaration]

[danglin at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125623

--- Comment #4 from John David Anglin <danglin at gcc dot gnu.org> ---
With gfortran-16, we get:

bash-5.1$ cat foo.f90.006t.original
__attribute__((fn spec (". r w w w ")))
void foo (real(kind=16) & restrict x, real(kind=16) & restrict f,
integer(kind=4) & restrict n, real(kind=16) & restrict y)
{
  integer(kind=4) m;

  {
    real(kind=16) D.1067;
    integer(kind=4) D.1068;

    D.1067 = *x;
    *f = !(ABS_EXPR <D.1067> u>
1.18973149535723176508575932662800701619646905264169404553e+4932) ?
__builtin_frexpl (D.1067, &D.1068) :  Nan;
  }
  {
    real(kind=16) D.1069;
    integer(kind=4) D.1070;

    D.1069 = *x;
    m = !(ABS_EXPR <D.1069> u>
1.18973149535723176508575932662800701619646905264169404553e+4932) ?
__builtin_frexpl (D.1069, &D.1070), D.1070 : 2147483647;
  }
  *y = __builtin_scalbnl (*f, 4);
}

The calls to __builtin_frexpl and __builtin_scalbnl eventually end
up as calls to frexpl and scalbnl, respectively.

I think you are correct that implements need to be added to c99_functions.c.
Messing with kinds.h doesn't fix this.