diff options
author | James Molloy <james.molloy@arm.com> | 2011-10-26 08:53:19 +0000 |
---|---|---|
committer | James Molloy <james.molloy@arm.com> | 2011-10-26 08:53:19 +0000 |
commit | dd9137aa56bd15e6db00eb9a652b10e76eb8c4fa (patch) | |
tree | 582dcf996da1914d448521479d38dc23b10735e7 /llvm/lib | |
parent | 39fe265e9ca879bd4495848f67394f48b6a9978d (diff) | |
download | bcm5719-llvm-dd9137aa56bd15e6db00eb9a652b10e76eb8c4fa.tar.gz bcm5719-llvm-dd9137aa56bd15e6db00eb9a652b10e76eb8c4fa.zip |
Revert r142530 at least temporarily while a discussion is had on llvm-commits regarding exactly how much optsize should optimize for size over performance.
llvm-svn: 143023
Diffstat (limited to 'llvm/lib')
-rw-r--r-- | llvm/lib/Target/ARM/ARMISelLowering.cpp | 7 |
1 files changed, 2 insertions, 5 deletions
diff --git a/llvm/lib/Target/ARM/ARMISelLowering.cpp b/llvm/lib/Target/ARM/ARMISelLowering.cpp index d7aca309807..472bf4c42f3 100644 --- a/llvm/lib/Target/ARM/ARMISelLowering.cpp +++ b/llvm/lib/Target/ARM/ARMISelLowering.cpp @@ -2104,11 +2104,8 @@ SDValue ARMTargetLowering::LowerGlobalAddressELF(SDValue Op, } // If we have T2 ops, we can materialize the address directly via movt/movw - // pair. This is always cheaper in terms of performance, but uses at least 2 - // extra bytes. - MachineFunction &MF = DAG.getMachineFunction(); - if (Subtarget->useMovt() && - !MF.getFunction()->hasFnAttr(Attribute::OptimizeForSize)) { + // pair. This is always cheaper. + if (Subtarget->useMovt()) { ++NumMovwMovt; // FIXME: Once remat is capable of dealing with instructions with register // operands, expand this into two nodes. |