diff options
| author | David Tweed <david.tweed@arm.com> | 2013-03-18 11:54:44 +0000 |
|---|---|---|
| committer | David Tweed <david.tweed@arm.com> | 2013-03-18 11:54:44 +0000 |
| commit | 5493feed251b00413a1978fc13a4a00fb7ec4e9e (patch) | |
| tree | c6c781fb550580e30f1520361909c0a13bb31a2a /llvm/lib/IR/DataLayout.cpp | |
| parent | 15bee188c4c141c3f82a86df3c3f4eba172c1706 (diff) | |
| download | bcm5719-llvm-5493feed251b00413a1978fc13a4a00fb7ec4e9e.tar.gz bcm5719-llvm-5493feed251b00413a1978fc13a4a00fb7ec4e9e.zip | |
The optimization a + (-0.0f) -> a was being misapplied to a + (+0.0f) in the vector case (because
we weren't differntiating floating-point zeroinitializers from other zero-initializers)
which was causing problems for code relying upon a + (+0.0f) to, eg, flush denormals to
0. Make the scalar and vector cases have the same behaviour.
llvm-svn: 177279
Diffstat (limited to 'llvm/lib/IR/DataLayout.cpp')
0 files changed, 0 insertions, 0 deletions
