]> git.ipfire.org Git - thirdparty/glibc.git/commit - math/math.h
Simplify C99 isgreater macros
authorWilco Dijkstra <wdijkstr@arm.com>
Thu, 28 Sep 2017 18:20:33 +0000 (19:20 +0100)
committerWilco Dijkstra <wdijkstr@arm.com>
Thu, 28 Sep 2017 18:43:54 +0000 (19:43 +0100)
commit1e6d07234fc0edcf0e88d75cf48f0b0dbbea3f39
tree55a43e82f97a2b808f13effeec49df332dfbc400
parentb2f03cf3a4b7ae4d1db155fba2180e3f580ce805
Simplify C99 isgreater macros

Simplify the C99 isgreater macros.  Although some support was added
in GCC 2.97, not all targets added support until GCC 3.1.  Therefore
only use the builtins in math.h from GCC 3.1 onwards, and defer to
generic macros otherwise.  Improve the generic isunordered macro
to use compares rather than call fpclassify twice - this is not only
faster but also correct for signaling NaNs.

* math/math.h: Improve handling of C99 isgreater macros.
* sysdeps/alpha/fpu/bits/mathinline.h: Remove isgreater macros.
* sysdeps/m68k/m680x0/fpu/bits/mathinline.h: Likewise.
* sysdeps/powerpc/bits/mathinline.h: Likewise.
* sysdeps/sparc/fpu/bits/mathinline.h: Likewise.
* sysdeps/x86/fpu/bits/mathinline.h: Likewise.
ChangeLog
math/math.h
sysdeps/alpha/fpu/bits/mathinline.h
sysdeps/m68k/m680x0/fpu/bits/mathinline.h
sysdeps/powerpc/bits/mathinline.h
sysdeps/sparc/fpu/bits/mathinline.h
sysdeps/x86/fpu/bits/mathinline.h