maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

AMDGPU/R600: Return correct chain when lowering loads 

The other LowerLOAD should be returning the correct chain.

llvm-svn: 241839
This commit is contained in:
Matt Arsenault 2015-07-09 18:47:03 +00:00
parent 6f062c8c2a
commit 8b03e6c164
1 changed files with 2 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
llvm/lib/Target/AMDGPU/R600ISelLowering.cpp
View File

@ -1460,14 +1460,8 @@ SDValue R600TargetLowering::LowerLOAD(SDValue Op, SelectionDAG &DAG) const
SDValue Ptr = Op.getOperand(1);
SDValue LoweredLoad;
SDValue Ret = AMDGPUTargetLowering::LowerLOAD(Op, DAG);
if (Ret.getNode()) {
SDValue Ops[2] = {
Ret,
Chain
};
return DAG.getMergeValues(Ops, DL);
}
if (SDValue Ret = AMDGPUTargetLowering::LowerLOAD(Op, DAG))
return Ret;
// Lower loads constant address space global variable loads
if (LoadNode->getAddressSpace() == AMDGPUAS::CONSTANT_ADDRESS &&