DAG: Undo and->or combine with FrameIndexes

This pattern shows up when lowering byval copies on AMDGPU.

The byval object access is split into 4-byte chunks, adding a
constant offset to the FixedStack base. When some of the offsets
turn into ors, this prevents combining the constant offsets.

This makes it not apparent that the object is there when matching
addressing modes, so it ends up using a scratch wave offset
relative access and the lengthy frame index expansion for that.

llvm-svn: 309775

1 files changed, 9 insertions, 0 deletions

diff --git a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index 54cddf77b4c..cf19d7b4fc0 100644
--- a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -1947,6 +1947,15 @@ SDValue DAGCombiner::visitADD(SDNode *N) {
         return DAG.getNode(ISD::ZERO_EXTEND, DL, VT, Not);
       }
     }
+
+    // Undo the add -> or combine to merge constant offsets from a frame index.
+    if (N0.getOpcode() == ISD::OR &&
+        isa<FrameIndexSDNode>(N0.getOperand(0)) &&
+        isa<ConstantSDNode>(N0.getOperand(1)) &&
+        DAG.haveNoCommonBitsSet(N0.getOperand(0), N0.getOperand(1))) {
+      SDValue Add0 = DAG.getNode(ISD::ADD, DL, VT, N1, N0.getOperand(1));
+      return DAG.getNode(ISD::ADD, DL, VT, N0.getOperand(0), Add0);
+    }
   }
 
   if (SDValue NewSel = foldBinOpIntoSelect(N))