Reimplement min_of_type using __MATH_TG.
commit219dd320d69deb9068f6b2ce46034d0eb4db888a
authorJoseph Myers <joseph@codesourcery.com>
Fri, 4 Aug 2017 21:32:57 +0000 (4 21:32 +0000)
committerJoseph Myers <joseph@codesourcery.com>
Fri, 4 Aug 2017 21:32:57 +0000 (4 21:32 +0000)
treeb882637bdd0e9b84688da36c352dd200115fec3e
parent64d4dea6cdcd908ea43a0eed5fc6262b1c6c205a
Reimplement min_of_type using __MATH_TG.

This patch reimplements the libm-internal min_of_type macro to use
__MATH_TG instead of its own local type-generic implementation, so
simplifying the code and reducing the number of different type-generic
implementation variants in use in glibc.

Tested for x86_64.

* sysdeps/generic/math_private.h (__EXPR_FLT128): Remove macro.
(min_of_type_f): New macro.
(min_of_type_): Likewise.
(min_of_type_l): Likewise.
(min_of_type_f128): Likewise.
(min_of_type): Define using __MATH_TG and taking an expression
argument.
(math_check_force_underflow): Pass expression instead of type to
min_of_type.
(math_check_force_underflow_nonneg): Likewise.
ChangeLog
sysdeps/generic/math_private.h