[PATCH 3/5] Make common fmax implementation generic.

[Paul E. Murphy]

On 08/31/2016 11:55 AM, Joseph Myers wrote:
> On Tue, 30 Aug 2016, Paul E. Murphy wrote:
> 
>> Also update aarch64 to ensure the correct s_fmin.c is included.
>> The include order favors including the generated copy.
>>
>> 	 * math/Makefile: (gen-libm-calls): Add s_fmax.
>> 	 (libm-calls): Remove above.
>>
>> 	 * math/s_fmax.c: Refactor into ...
>> 	 * math/s_fmax_template.c: New file.
>> 	 * math/s_fmaxf.c: Removed.
>> 	 * math/s_fmaxl.c: Removed.
>>
>> 	 * sysdeps/ieee754/ldbl-opt/math-type-macros-double.h:
>> 	 (LDOUBLE_fmaxl_libm_version): New macro.
>>
>> 	 * sysdeps/ieee754/ldbl-opt/s_fmax.c: Removed.
>> 	 * sysdeps/ieee754/ldbl-opt/s_fmaxl.c: Removed.
>>
>> 	* sysdeps/aarch64/fpu/s_fmax.c: Append fpu/ to include of
>> 	s_fmin.c to pick the aarch64 version of the file.
>> 	* sysdeps/aarch64/fpu/s_fmaxf.c: Likewise.
>> 	* sysdeps/aarch64/fpu/s_fminf.c: Likewise.
> 
> OK.
> 

Committed as 847c916.