[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]

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

John David Anglin <danglin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #64638|0                           |1
        is obsolete|                            |

--- Comment #9 from John David Anglin <danglin at gcc dot gnu.org> ---
Created attachment 64653
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=64653&action=edit
Implement strnlen, frexpl and scalbnl c99 functions

Noticed strnlen was missing from testsuite results.

frexpl and scalbnl implementations are derived from libquadmath.  Currently,
I only implemented 128-bit long double.  Glibc has 80 bit if needed but
licensing differs.

Warning compiling copysignl is bogus.  There is no recursion in generated code.

Added new protos.  This fixes libgfortran build error.

We need to include string.h for memchr() used in strnlen() implementation.