Prevent a DAGCombine from firing where there are two uses of
a combined-away node and the result of the combine isn't substantially
smaller than the input, it's just canonicalized. This is the first part
of a significant (7%) performance gain for Snappy's hot decompression
loop.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@147604 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index 8a8f0cf..5b2d86f 100644
--- a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -3331,7 +3331,9 @@
 
   // fold (shl (srl x, c1), c2) -> (and (shl x, (sub c2, c1), MASK) or
   //                               (and (srl x, (sub c1, c2), MASK)
-  if (N1C && N0.getOpcode() == ISD::SRL &&
+  // Only fold this if the inner shift has no other uses -- if it does, folding
+  // this will increase the total number of instructions.
+  if (N1C && N0.getOpcode() == ISD::SRL && N0.hasOneUse() &&
       N0.getOperand(1).getOpcode() == ISD::Constant) {
     uint64_t c1 = cast<ConstantSDNode>(N0.getOperand(1))->getZExtValue();
     if (c1 < VT.getSizeInBits()) {