Index: include/llvm/ADT/DepthFirstIterator.h
===================================================================
--- include/llvm/ADT/DepthFirstIterator.h
+++ include/llvm/ADT/DepthFirstIterator.h
@@ -58,10 +58,25 @@
   SetType &Visited;
 };
 
+// The visited stated for the iteration is a simple set augmented with
+// one more method, completed, which is invoked when all children of a
+// node have been processed. It is intended to distinguish of back and
+// cross edges in the spanning tree but is not used in the common case.
+template <typename NodeRef, unsigned SmallSize=8>
+struct df_iterator_default_set : public llvm::SmallPtrSet<NodeRef, SmallSize> {
+  typedef llvm::SmallPtrSet<NodeRef, SmallSize>  BaseSet;
+  typedef typename BaseSet::iterator iterator;
+  std::pair<iterator,bool> insert(NodeRef N) { return BaseSet::insert(N) ; }
+  template <typename IterT>
+  void insert(IterT Begin, IterT End) { BaseSet::insert(Begin,End); }
+
+  void completed(NodeRef) { }
+};
+
 // Generic Depth First Iterator
 template <class GraphT,
           class SetType =
-              llvm::SmallPtrSet<typename GraphTraits<GraphT>::NodeRef, 8>,
+              df_iterator_default_set<typename GraphTraits<GraphT>::NodeRef>,
           bool ExtStorage = false, class GT = GraphTraits<GraphT>>
 class df_iterator
     : public std::iterator<std::forward_iterator_tag, typename GT::NodeRef>,
@@ -89,10 +104,8 @@
   }
   inline df_iterator(NodeRef Node, SetType &S)
       : df_iterator_storage<SetType, ExtStorage>(S) {
-    if (!S.count(Node)) {
+    if (this->Visited.insert(Node).second)
       VisitStack.push_back(StackElement(Node, None));
-      this->Visited.insert(Node);
-    }
   }
   inline df_iterator(SetType &S)
     : df_iterator_storage<SetType, ExtStorage>(S) {
@@ -119,7 +132,8 @@
           return;
         }
       }
-
+      this->Visited.completed(Node);
+      
       // Oops, ran out of successors... go up a level on the stack.
       VisitStack.pop_back();
     } while (!VisitStack.empty());
@@ -235,7 +249,8 @@
 
 // Provide global definitions of inverse depth first iterators...
 template <class T,
-          class SetTy = llvm::SmallPtrSet<typename GraphTraits<T>::NodeRef, 8>,
+          class SetTy =
+              df_iterator_default_set<typename GraphTraits<T>::NodeRef>,
           bool External = false>
 struct idf_iterator : public df_iterator<Inverse<T>, SetTy, External> {
   idf_iterator(const df_iterator<Inverse<T>, SetTy, External> &V)
Index: include/llvm/Analysis/LoopInfoImpl.h
===================================================================
--- include/llvm/Analysis/LoopInfoImpl.h
+++ include/llvm/Analysis/LoopInfoImpl.h
@@ -228,9 +228,9 @@
   // Setup for using a depth-first iterator to visit every block in the loop.
   SmallVector<BlockT*, 8> ExitBBs;
   getExitBlocks(ExitBBs);
-  llvm::SmallPtrSet<BlockT*, 8> VisitSet;
+  df_iterator_default_set<BlockT*> VisitSet;
   VisitSet.insert(ExitBBs.begin(), ExitBBs.end());
-  df_ext_iterator<BlockT*, llvm::SmallPtrSet<BlockT*, 8> >
+  df_ext_iterator<BlockT*, df_iterator_default_set<BlockT*>>
     BI = df_ext_begin(getHeader(), VisitSet),
     BE = df_ext_end(getHeader(), VisitSet);
 
Index: include/llvm/Analysis/RegionInfo.h
===================================================================
--- include/llvm/Analysis/RegionInfo.h
+++ include/llvm/Analysis/RegionInfo.h
@@ -626,12 +626,14 @@
   /// are direct children of this Region. It does not iterate over any
   /// RegionNodes that are also element of a subregion of this Region.
   //@{
-  typedef df_iterator<RegionNodeT *, SmallPtrSet<RegionNodeT *, 8>, false,
-                      GraphTraits<RegionNodeT *>> element_iterator;
-
-  typedef df_iterator<const RegionNodeT *, SmallPtrSet<const RegionNodeT *, 8>,
-                      false,
-                      GraphTraits<const RegionNodeT *>> const_element_iterator;
+  typedef df_iterator<RegionNodeT *, df_iterator_default_set<RegionNodeT *>,
+                      false, GraphTraits<RegionNodeT *>>
+      element_iterator;
+
+  typedef df_iterator<const RegionNodeT *,
+                      df_iterator_default_set<const RegionNodeT *>, false,
+                      GraphTraits<const RegionNodeT *>>
+      const_element_iterator;
 
   element_iterator element_begin();
   element_iterator element_end();
Index: include/llvm/Analysis/RegionIterator.h
===================================================================
--- include/llvm/Analysis/RegionIterator.h
+++ include/llvm/Analysis/RegionIterator.h
@@ -294,7 +294,7 @@
   template <>                                                                  \
   struct GraphTraits<FlatIt<RegionT *>>                                        \
       : public GraphTraits<FlatIt<NodeT *>> {                                  \
-    typedef df_iterator<NodeRef, SmallPtrSet<NodeRef, 8>, false,               \
+    typedef df_iterator<NodeRef, df_iterator_default_set<NodeRef>, false,      \
                         GraphTraits<FlatIt<NodeRef>>>                          \
         nodes_iterator;                                                        \
     static NodeRef getEntryNode(RegionT *R) {                                  \
@@ -316,7 +316,7 @@
 
 template <> struct GraphTraits<RegionInfo*>
   : public GraphTraits<FlatIt<RegionNode*> > {
-  typedef df_iterator<NodeRef, SmallPtrSet<NodeRef, 8>, false,
+  typedef df_iterator<NodeRef, df_iterator_default_set<NodeRef>, false,
                       GraphTraits<FlatIt<NodeRef>>>
       nodes_iterator;
 
@@ -333,7 +333,7 @@
 
 template <> struct GraphTraits<RegionInfoPass*>
   : public GraphTraits<RegionInfo *> {
-  typedef df_iterator<NodeRef, SmallPtrSet<NodeRef, 8>, false,
+  typedef df_iterator<NodeRef, df_iterator_default_set<NodeRef>, false,
                       GraphTraits<FlatIt<NodeRef>>>
       nodes_iterator;
 
Index: include/llvm/CodeGen/MachineRegionInfo.h
===================================================================
--- include/llvm/CodeGen/MachineRegionInfo.h
+++ include/llvm/CodeGen/MachineRegionInfo.h
@@ -142,7 +142,7 @@
 
 template <> struct GraphTraits<MachineRegionInfo*>
   : public GraphTraits<FlatIt<MachineRegionNode*> > {
-  typedef df_iterator<NodeRef, SmallPtrSet<NodeRef, 8>, false,
+  typedef df_iterator<NodeRef, df_iterator_default_set<NodeRef>, false,
                       GraphTraits<FlatIt<NodeRef>>>
       nodes_iterator;
 
@@ -159,7 +159,7 @@
 
 template <> struct GraphTraits<MachineRegionInfoPass*>
   : public GraphTraits<MachineRegionInfo *> {
-  typedef df_iterator<NodeRef, SmallPtrSet<NodeRef, 8>, false,
+  typedef df_iterator<NodeRef, df_iterator_default_set<NodeRef>, false,
                       GraphTraits<FlatIt<NodeRef>>>
       nodes_iterator;
 
Index: include/llvm/IR/Dominators.h
===================================================================
--- include/llvm/IR/Dominators.h
+++ include/llvm/IR/Dominators.h
@@ -157,7 +157,7 @@
 template <class Node, class ChildIterator> struct DomTreeGraphTraitsBase {
   typedef Node *NodeRef;
   typedef ChildIterator ChildIteratorType;
-  typedef df_iterator<Node *, SmallPtrSet<NodeRef, 8>> nodes_iterator;
+  typedef df_iterator<Node *, df_iterator_default_set<Node*>> nodes_iterator;
 
   static NodeRef getEntryNode(NodeRef N) { return N; }
   static ChildIteratorType child_begin(NodeRef N) { return N->begin(); }
Index: lib/CodeGen/LiveIntervalAnalysis.cpp
===================================================================
--- lib/CodeGen/LiveIntervalAnalysis.cpp
+++ lib/CodeGen/LiveIntervalAnalysis.cpp
@@ -599,7 +599,7 @@
   // Find all blocks that are reachable from KillMBB without leaving VNI's live
   // range. It is possible that KillMBB itself is reachable, so start a DFS
   // from each successor.
-  typedef SmallPtrSet<MachineBasicBlock*, 9> VisitedTy;
+  typedef df_iterator_default_set<MachineBasicBlock*,9> VisitedTy;
   VisitedTy Visited;
   for (MachineBasicBlock::succ_iterator
        SuccI = KillMBB->succ_begin(), SuccE = KillMBB->succ_end();
Index: lib/CodeGen/LiveVariables.cpp
===================================================================
--- lib/CodeGen/LiveVariables.cpp
+++ lib/CodeGen/LiveVariables.cpp
@@ -643,7 +643,7 @@
   // register before its uses due to dominance properties of SSA (except for PHI
   // nodes, which are treated as a special case).
   MachineBasicBlock *Entry = &MF->front();
-  SmallPtrSet<MachineBasicBlock*,16> Visited;
+  df_iterator_default_set<MachineBasicBlock*,16> Visited;
 
   for (MachineBasicBlock *MBB : depth_first_ext(Entry, Visited)) {
     runOnBlock(MBB, NumRegs);
Index: lib/CodeGen/MachineVerifier.cpp
===================================================================
--- lib/CodeGen/MachineVerifier.cpp
+++ lib/CodeGen/MachineVerifier.cpp
@@ -2014,11 +2014,11 @@
 
   SmallVector<StackStateOfBB, 8> SPState;
   SPState.resize(MF->getNumBlockIDs());
-  SmallPtrSet<const MachineBasicBlock*, 8> Reachable;
+  df_iterator_default_set<const MachineBasicBlock*> Reachable;
 
   // Visit the MBBs in DFS order.
   for (df_ext_iterator<const MachineFunction*,
-                       SmallPtrSet<const MachineBasicBlock*, 8> >
+                       df_iterator_default_set<const MachineBasicBlock*> >
        DFI = df_ext_begin(MF, Reachable), DFE = df_ext_end(MF, Reachable);
        DFI != DFE; ++DFI) {
     const MachineBasicBlock *MBB = *DFI;
Index: lib/CodeGen/PrologEpilogInserter.cpp
===================================================================
--- lib/CodeGen/PrologEpilogInserter.cpp
+++ lib/CodeGen/PrologEpilogInserter.cpp
@@ -1008,7 +1008,7 @@
   // Store SPAdj at exit of a basic block.
   SmallVector<int, 8> SPState;
   SPState.resize(Fn.getNumBlockIDs());
-  SmallPtrSet<MachineBasicBlock*, 8> Reachable;
+  df_iterator_default_set<MachineBasicBlock*> Reachable;
 
   // Iterate over the reachable blocks in DFS order.
   for (auto DFI = df_ext_begin(&Fn, Reachable), DFE = df_ext_end(&Fn, Reachable);
Index: lib/CodeGen/UnreachableBlockElim.cpp
===================================================================
--- lib/CodeGen/UnreachableBlockElim.cpp
+++ lib/CodeGen/UnreachableBlockElim.cpp
@@ -40,7 +40,7 @@
 using namespace llvm;
 
 static bool eliminateUnreachableBlock(Function &F) {
-  SmallPtrSet<BasicBlock*, 8> Reachable;
+  df_iterator_default_set<BasicBlock*> Reachable;
 
   // Mark all reachable blocks.
   for (BasicBlock *BB : depth_first_ext(&F, Reachable))
@@ -130,7 +130,7 @@
 }
 
 bool UnreachableMachineBlockElim::runOnMachineFunction(MachineFunction &F) {
-  SmallPtrSet<MachineBasicBlock*, 8> Reachable;
+  df_iterator_default_set<MachineBasicBlock*> Reachable;
   bool ModifiedPHI = false;
 
   MMI = getAnalysisIfAvailable<MachineModuleInfo>();
Index: lib/Target/X86/X86FloatingPoint.cpp
===================================================================
--- lib/Target/X86/X86FloatingPoint.cpp
+++ lib/Target/X86/X86FloatingPoint.cpp
@@ -326,7 +326,7 @@
 
   // Process the function in depth first order so that we process at least one
   // of the predecessors for every reachable block in the function.
-  SmallPtrSet<MachineBasicBlock*, 8> Processed;
+  df_iterator_default_set<MachineBasicBlock*> Processed;
   MachineBasicBlock *Entry = &MF.front();
 
   bool Changed = false;
Index: lib/Transforms/IPO/ArgumentPromotion.cpp
===================================================================
--- lib/Transforms/IPO/ArgumentPromotion.cpp
+++ lib/Transforms/IPO/ArgumentPromotion.cpp
@@ -600,7 +600,7 @@
 
   // Because there could be several/many load instructions, remember which
   // blocks we know to be transparent to the load.
-  SmallPtrSet<BasicBlock*, 16> TranspBlocks;
+  df_iterator_default_set<BasicBlock*, 16> TranspBlocks;
 
   for (LoadInst *Load : Loads) {
     // Check to see if the load is invalidated from the start of the block to
Index: unittests/ADT/DepthFirstIteratorTest.cpp
===================================================================
--- unittests/ADT/DepthFirstIteratorTest.cpp
+++ unittests/ADT/DepthFirstIteratorTest.cpp
@@ -27,6 +27,8 @@
   }
 
   size_t count(const T &Item) const { return S.count(Item); }
+  
+  void completed(T) { }
 };
 
 template <typename T> class df_iterator_storage<CountedSet<T>, true> {