This adds new options -fdenormal-fp-math and passes through option -ffast-math
authorSjoerd Meijer <sjoerd.meijer@arm.com>
Tue, 30 Aug 2016 08:09:45 +0000 (08:09 +0000)
committerSjoerd Meijer <sjoerd.meijer@arm.com>
Tue, 30 Aug 2016 08:09:45 +0000 (08:09 +0000)
commit0a8d4216ad52aeffc14597a2d946134afa0d89ec
treea93cd5532ced3f54265fdc9fd20b9585d63363e9
parent5a99207d11f3ea72ca9424596e1d0cdf4bdb9f3f
This adds new options -fdenormal-fp-math and passes through option -ffast-math
to CC1, which are translated to function attributes and can e.g. be mapped on
build attributes FP_exceptions and FP_denormal. Setting these build attributes
allows better selection of floating point libraries.

Differential Revision: https://reviews.llvm.org/D23840

llvm-svn: 280064
clang/docs/UsersManual.rst
clang/include/clang/Driver/Options.td
clang/include/clang/Frontend/CodeGenOptions.def
clang/include/clang/Frontend/CodeGenOptions.h
clang/lib/CodeGen/CGCall.cpp
clang/lib/Driver/Tools.cpp
clang/lib/Frontend/CompilerInvocation.cpp
clang/test/CodeGen/denormalfpmode.c [new file with mode: 0644]
clang/test/CodeGen/noexceptionsfpmath.c [new file with mode: 0644]
clang/test/Driver/fast-math.c