diff options
| author | Hans Wennborg <hans@hanshq.net> | 2016-03-30 23:38:01 +0000 |
|---|---|---|
| committer | Hans Wennborg <hans@hanshq.net> | 2016-03-30 23:38:01 +0000 |
| commit | 6596977130f9fb5179b16d3c3c291c69efd627bb (patch) | |
| tree | 947efc1f13ea3ee726a20057903113f151c3dd88 /llvm/lib/Target/X86/X86CallFrameOptimization.cpp | |
| parent | 0e450a5d61795a2ca9160a34a75855fd1d16e16c (diff) | |
| download | llvm-6596977130f9fb5179b16d3c3c291c69efd627bb.zip llvm-6596977130f9fb5179b16d3c3c291c69efd627bb.tar.gz llvm-6596977130f9fb5179b16d3c3c291c69efd627bb.tar.bz2 | |
[X86] Enable call frame optimization ("mov to push") not only for optsize (PR26325)
The size savings are significant, and from what I can tell, both ICC and GCC do this.
Differential Revision: http://reviews.llvm.org/D18573
llvm-svn: 264966
Diffstat (limited to 'llvm/lib/Target/X86/X86CallFrameOptimization.cpp')
| -rw-r--r-- | llvm/lib/Target/X86/X86CallFrameOptimization.cpp | 4 |
1 files changed, 0 insertions, 4 deletions
diff --git a/llvm/lib/Target/X86/X86CallFrameOptimization.cpp b/llvm/lib/Target/X86/X86CallFrameOptimization.cpp index e86b5d4..1ba45ff 100644 --- a/llvm/lib/Target/X86/X86CallFrameOptimization.cpp +++ b/llvm/lib/Target/X86/X86CallFrameOptimization.cpp @@ -180,10 +180,6 @@ bool X86CallFrameOptimization::isProfitable(MachineFunction &MF, if (CannotReserveFrame) return true; - // Don't do this when not optimizing for size. - if (!MF.getFunction()->optForSize()) - return false; - unsigned StackAlign = TFL->getStackAlignment(); int64_t Advantage = 0; |