diff options
| author | Sanjay Patel <spatel@rotateright.com> | 2015-08-05 15:12:03 +0000 |
|---|---|---|
| committer | Sanjay Patel <spatel@rotateright.com> | 2015-08-05 15:12:03 +0000 |
| commit | b6a79f9916f6b676e2ad250b9e125d9f7b165fe5 (patch) | |
| tree | f3b9d8e3f2ed18a3e0e1b17e43a08f95b3d4f2dc /llvm/lib/CodeGen/MachineSSAUpdater.cpp | |
| parent | fc29f69d74fbb3a585d3df7dd2c7cef290c7a2be (diff) | |
| download | bcm5719-llvm-b6a79f9916f6b676e2ad250b9e125d9f7b165fe5.tar.gz bcm5719-llvm-b6a79f9916f6b676e2ad250b9e125d9f7b165fe5.zip | |
revert r243687: enable fast-math-flag propagation to DAG nodes
We can't propagate FMF partially without breaking DAG-level CSE. We either need to
relax CSE to account for mismatched FMF as a temporary work-around or fully propagate
FMF throughout the DAG.
Surprisingly, there are no existing regression tests for this, but here's an example:
define float @fmf(float %a, float %b) {
%mul1 = fmul fast float %a, %b
%nega = fsub fast float 0.0, %a
%mul2 = fmul fast float %nega, %b
%abx2 = fsub fast float %mul1, %mul2
ret float %abx2
}
$ llc -o - badflags.ll -march=x86-64 -mattr=fma -enable-unsafe-fp-math -enable-fmf-dag=0
...
vmulss %xmm1, %xmm0, %xmm0
vaddss %xmm0, %xmm0, %xmm0
retq
$ llc -o - badflags.ll -march=x86-64 -mattr=fma -enable-unsafe-fp-math -enable-fmf-dag=1
...
vmulss %xmm1, %xmm0, %xmm2
vfmadd213ss %xmm2, %xmm1, %xmm0 <--- failed to recognize that (a * b) was already calculated
retq
llvm-svn: 244053
Diffstat (limited to 'llvm/lib/CodeGen/MachineSSAUpdater.cpp')
0 files changed, 0 insertions, 0 deletions

