Import LLVM, at r72732.

7 files changed, 4091 insertions, 0 deletions

diff --git a/lib/Analysis/IPA/Andersens.cpp b/lib/Analysis/IPA/Andersens.cpp
new file mode 100644
index 0000000..8584d06
--- /dev/null
+++ b/lib/Analysis/IPA/Andersens.cpp
@@ -0,0 +1,2878 @@
+//===- Andersens.cpp - Andersen's Interprocedural Alias Analysis ----------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This file defines an implementation of Andersen's interprocedural alias
+// analysis
+//
+// In pointer analysis terms, this is a subset-based, flow-insensitive,
+// field-sensitive, and context-insensitive algorithm pointer algorithm.
+//
+// This algorithm is implemented as three stages:
+//   1. Object identification.
+//   2. Inclusion constraint identification.
+//   3. Offline constraint graph optimization
+//   4. Inclusion constraint solving.
+//
+// The object identification stage identifies all of the memory objects in the
+// program, which includes globals, heap allocated objects, and stack allocated
+// objects.
+//
+// The inclusion constraint identification stage finds all inclusion constraints
+// in the program by scanning the program, looking for pointer assignments and
+// other statements that effect the points-to graph.  For a statement like "A =
+// B", this statement is processed to indicate that A can point to anything that
+// B can point to.  Constraints can handle copies, loads, and stores, and
+// address taking.
+//
+// The offline constraint graph optimization portion includes offline variable
+// substitution algorithms intended to compute pointer and location
+// equivalences.  Pointer equivalences are those pointers that will have the
+// same points-to sets, and location equivalences are those variables that
+// always appear together in points-to sets.  It also includes an offline
+// cycle detection algorithm that allows cycles to be collapsed sooner 
+// during solving.
+//
+// The inclusion constraint solving phase iteratively propagates the inclusion
+// constraints until a fixed point is reached.  This is an O(N^3) algorithm.
+//
+// Function constraints are handled as if they were structs with X fields.
+// Thus, an access to argument X of function Y is an access to node index
+// getNode(Y) + X.  This representation allows handling of indirect calls
+// without any issues.  To wit, an indirect call Y(a,b) is equivalent to
+// *(Y + 1) = a, *(Y + 2) = b.
+// The return node for a function is always located at getNode(F) +
+// CallReturnPos. The arguments start at getNode(F) + CallArgPos.
+//
+// Future Improvements:
+//   Use of BDD's.
+//===----------------------------------------------------------------------===//
+
+#define DEBUG_TYPE "anders-aa"
+#include "llvm/Constants.h"
+#include "llvm/DerivedTypes.h"
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+#include "llvm/Support/Compiler.h"
+#include "llvm/Support/InstIterator.h"
+#include "llvm/Support/InstVisitor.h"
+#include "llvm/Analysis/AliasAnalysis.h"
+#include "llvm/Analysis/Passes.h"
+#include "llvm/Support/Debug.h"
+#include "llvm/ADT/Statistic.h"
+#include "llvm/ADT/SparseBitVector.h"
+#include "llvm/ADT/DenseSet.h"
+#include <algorithm>
+#include <set>
+#include <list>
+#include <map>
+#include <stack>
+#include <vector>
+#include <queue>
+
+// Determining the actual set of nodes the universal set can consist of is very
+// expensive because it means propagating around very large sets.  We rely on
+// other analysis being able to determine which nodes can never be pointed to in
+// order to disambiguate further than "points-to anything".
+#define FULL_UNIVERSAL 0
+
+using namespace llvm;
+STATISTIC(NumIters      , "Number of iterations to reach convergence");
+STATISTIC(NumConstraints, "Number of constraints");
+STATISTIC(NumNodes      , "Number of nodes");
+STATISTIC(NumUnified    , "Number of variables unified");
+STATISTIC(NumErased     , "Number of redundant constraints erased");
+
+static const unsigned SelfRep = (unsigned)-1;
+static const unsigned Unvisited = (unsigned)-1;
+// Position of the function return node relative to the function node.
+static const unsigned CallReturnPos = 1;
+// Position of the function call node relative to the function node.
+static const unsigned CallFirstArgPos = 2;
+
+namespace {
+  struct BitmapKeyInfo {
+    static inline SparseBitVector<> *getEmptyKey() {
+      return reinterpret_cast<SparseBitVector<> *>(-1);
+    }
+    static inline SparseBitVector<> *getTombstoneKey() {
+      return reinterpret_cast<SparseBitVector<> *>(-2);
+    }
+    static unsigned getHashValue(const SparseBitVector<> *bitmap) {
+      return bitmap->getHashValue();
+    }
+    static bool isEqual(const SparseBitVector<> *LHS,
+                        const SparseBitVector<> *RHS) {
+      if (LHS == RHS)
+        return true;
+      else if (LHS == getEmptyKey() || RHS == getEmptyKey()
+               || LHS == getTombstoneKey() || RHS == getTombstoneKey())
+        return false;
+
+      return *LHS == *RHS;
+    }
+
+    static bool isPod() { return true; }
+  };
+
+  class VISIBILITY_HIDDEN Andersens : public ModulePass, public AliasAnalysis,
+                                      private InstVisitor<Andersens> {
+    struct Node;
+
+    /// Constraint - Objects of this structure are used to represent the various
+    /// constraints identified by the algorithm.  The constraints are 'copy',
+    /// for statements like "A = B", 'load' for statements like "A = *B",
+    /// 'store' for statements like "*A = B", and AddressOf for statements like
+    /// A = alloca;  The Offset is applied as *(A + K) = B for stores,
+    /// A = *(B + K) for loads, and A = B + K for copies.  It is
+    /// illegal on addressof constraints (because it is statically
+    /// resolvable to A = &C where C = B + K)
+
+    struct Constraint {
+      enum ConstraintType { Copy, Load, Store, AddressOf } Type;
+      unsigned Dest;
+      unsigned Src;
+      unsigned Offset;
+
+      Constraint(ConstraintType Ty, unsigned D, unsigned S, unsigned O = 0)
+        : Type(Ty), Dest(D), Src(S), Offset(O) {
+        assert((Offset == 0 || Ty != AddressOf) &&
+               "Offset is illegal on addressof constraints");
+      }
+
+      bool operator==(const Constraint &RHS) const {
+        return RHS.Type == Type
+          && RHS.Dest == Dest
+          && RHS.Src == Src
+          && RHS.Offset == Offset;
+      }
+
+      bool operator!=(const Constraint &RHS) const {
+        return !(*this == RHS);
+      }
+
+      bool operator<(const Constraint &RHS) const {
+        if (RHS.Type != Type)
+          return RHS.Type < Type;
+        else if (RHS.Dest != Dest)
+          return RHS.Dest < Dest;
+        else if (RHS.Src != Src)
+          return RHS.Src < Src;
+        return RHS.Offset < Offset;
+      }
+    };
+
+    // Information DenseSet requires implemented in order to be able to do
+    // it's thing
+    struct PairKeyInfo {
+      static inline std::pair<unsigned, unsigned> getEmptyKey() {
+        return std::make_pair(~0U, ~0U);
+      }
+      static inline std::pair<unsigned, unsigned> getTombstoneKey() {
+        return std::make_pair(~0U - 1, ~0U - 1);
+      }
+      static unsigned getHashValue(const std::pair<unsigned, unsigned> &P) {
+        return P.first ^ P.second;
+      }
+      static unsigned isEqual(const std::pair<unsigned, unsigned> &LHS,
+                              const std::pair<unsigned, unsigned> &RHS) {
+        return LHS == RHS;
+      }
+    };
+    
+    struct ConstraintKeyInfo {
+      static inline Constraint getEmptyKey() {
+        return Constraint(Constraint::Copy, ~0U, ~0U, ~0U);
+      }
+      static inline Constraint getTombstoneKey() {
+        return Constraint(Constraint::Copy, ~0U - 1, ~0U - 1, ~0U - 1);
+      }
+      static unsigned getHashValue(const Constraint &C) {
+        return C.Src ^ C.Dest ^ C.Type ^ C.Offset;
+      }
+      static bool isEqual(const Constraint &LHS,
+                          const Constraint &RHS) {
+        return LHS.Type == RHS.Type && LHS.Dest == RHS.Dest
+          && LHS.Src == RHS.Src && LHS.Offset == RHS.Offset;
+      }
+    };
+
+    // Node class - This class is used to represent a node in the constraint
+    // graph.  Due to various optimizations, it is not always the case that
+    // there is a mapping from a Node to a Value.  In particular, we add
+    // artificial Node's that represent the set of pointed-to variables shared
+    // for each location equivalent Node.
+    struct Node {
+    private:
+      static unsigned Counter;
+
+    public:
+      Value *Val;
+      SparseBitVector<> *Edges;
+      SparseBitVector<> *PointsTo;
+      SparseBitVector<> *OldPointsTo;
+      std::list<Constraint> Constraints;
+
+      // Pointer and location equivalence labels
+      unsigned PointerEquivLabel;
+      unsigned LocationEquivLabel;
+      // Predecessor edges, both real and implicit
+      SparseBitVector<> *PredEdges;
+      SparseBitVector<> *ImplicitPredEdges;
+      // Set of nodes that point to us, only use for location equivalence.
+      SparseBitVector<> *PointedToBy;
+      // Number of incoming edges, used during variable substitution to early
+      // free the points-to sets
+      unsigned NumInEdges;
+      // True if our points-to set is in the Set2PEClass map
+      bool StoredInHash;
+      // True if our node has no indirect constraints (complex or otherwise)
+      bool Direct;
+      // True if the node is address taken, *or* it is part of a group of nodes
+      // that must be kept together.  This is set to true for functions and
+      // their arg nodes, which must be kept at the same position relative to
+      // their base function node.
+      bool AddressTaken;
+
+      // Nodes in cycles (or in equivalence classes) are united together using a
+      // standard union-find representation with path compression.  NodeRep
+      // gives the index into GraphNodes for the representative Node.
+      unsigned NodeRep;
+
+      // Modification timestamp.  Assigned from Counter.
+      // Used for work list prioritization.
+      unsigned Timestamp;
+
+      explicit Node(bool direct = true) :
+        Val(0), Edges(0), PointsTo(0), OldPointsTo(0), 
+        PointerEquivLabel(0), LocationEquivLabel(0), PredEdges(0),
+        ImplicitPredEdges(0), PointedToBy(0), NumInEdges(0),
+        StoredInHash(false), Direct(direct), AddressTaken(false),
+        NodeRep(SelfRep), Timestamp(0) { }
+
+      Node *setValue(Value *V) {
+        assert(Val == 0 && "Value already set for this node!");
+        Val = V;
+        return this;
+      }
+
+      /// getValue - Return the LLVM value corresponding to this node.
+      ///
+      Value *getValue() const { return Val; }
+
+      /// addPointerTo - Add a pointer to the list of pointees of this node,
+      /// returning true if this caused a new pointer to be added, or false if
+      /// we already knew about the points-to relation.
+      bool addPointerTo(unsigned Node) {
+        return PointsTo->test_and_set(Node);
+      }
+
+      /// intersects - Return true if the points-to set of this node intersects
+      /// with the points-to set of the specified node.
+      bool intersects(Node *N) const;
+
+      /// intersectsIgnoring - Return true if the points-to set of this node
+      /// intersects with the points-to set of the specified node on any nodes
+      /// except for the specified node to ignore.
+      bool intersectsIgnoring(Node *N, unsigned) const;
+
+      // Timestamp a node (used for work list prioritization)
+      void Stamp() {
+        Timestamp = Counter++;
+      }
+
+      bool isRep() const {
+        return( (int) NodeRep < 0 );
+      }
+    };
+
+    struct WorkListElement {
+      Node* node;
+      unsigned Timestamp;
+      WorkListElement(Node* n, unsigned t) : node(n), Timestamp(t) {}
+
+      // Note that we reverse the sense of the comparison because we
+      // actually want to give low timestamps the priority over high,
+      // whereas priority is typically interpreted as a greater value is
+      // given high priority.
+      bool operator<(const WorkListElement& that) const {
+        return( this->Timestamp > that.Timestamp );
+      }
+    };
+
+    // Priority-queue based work list specialized for Nodes.
+    class WorkList {
+      std::priority_queue<WorkListElement> Q;
+
+    public:
+      void insert(Node* n) {
+        Q.push( WorkListElement(n, n->Timestamp) );
+      }
+
+      // We automatically discard non-representative nodes and nodes
+      // that were in the work list twice (we keep a copy of the
+      // timestamp in the work list so we can detect this situation by
+      // comparing against the node's current timestamp).
+      Node* pop() {
+        while( !Q.empty() ) {
+          WorkListElement x = Q.top(); Q.pop();
+          Node* INode = x.node;
+
+          if( INode->isRep() &&
+              INode->Timestamp == x.Timestamp ) {
+            return(x.node);
+          }
+        }
+        return(0);
+      }
+
+      bool empty() {
+        return Q.empty();
+      }
+    };
+
+    /// GraphNodes - This vector is populated as part of the object
+    /// identification stage of the analysis, which populates this vector with a
+    /// node for each memory object and fills in the ValueNodes map.
+    std::vector<Node> GraphNodes;
+
+    /// ValueNodes - This map indicates the Node that a particular Value* is
+    /// represented by.  This contains entries for all pointers.
+    DenseMap<Value*, unsigned> ValueNodes;
+
+    /// ObjectNodes - This map contains entries for each memory object in the
+    /// program: globals, alloca's and mallocs.
+    DenseMap<Value*, unsigned> ObjectNodes;
+
+    /// ReturnNodes - This map contains an entry for each function in the
+    /// program that returns a value.
+    DenseMap<Function*, unsigned> ReturnNodes;
+
+    /// VarargNodes - This map contains the entry used to represent all pointers
+    /// passed through the varargs portion of a function call for a particular
+    /// function.  An entry is not present in this map for functions that do not
+    /// take variable arguments.
+    DenseMap<Function*, unsigned> VarargNodes;
+
+
+    /// Constraints - This vector contains a list of all of the constraints
+    /// identified by the program.
+    std::vector<Constraint> Constraints;
+
+    // Map from graph node to maximum K value that is allowed (for functions,
+    // this is equivalent to the number of arguments + CallFirstArgPos)
+    std::map<unsigned, unsigned> MaxK;
+
+    /// This enum defines the GraphNodes indices that correspond to important
+    /// fixed sets.
+    enum {
+      UniversalSet = 0,
+      NullPtr      = 1,
+      NullObject   = 2,
+      NumberSpecialNodes
+    };
+    // Stack for Tarjan's
+    std::stack<unsigned> SCCStack;
+    // Map from Graph Node to DFS number
+    std::vector<unsigned> Node2DFS;
+    // Map from Graph Node to Deleted from graph.
+    std::vector<bool> Node2Deleted;
+    // Same as Node Maps, but implemented as std::map because it is faster to
+    // clear 
+    std::map<unsigned, unsigned> Tarjan2DFS;
+    std::map<unsigned, bool> Tarjan2Deleted;
+    // Current DFS number
+    unsigned DFSNumber;
+
+    // Work lists.
+    WorkList w1, w2;
+    WorkList *CurrWL, *NextWL; // "current" and "next" work lists
+
+    // Offline variable substitution related things
+
+    // Temporary rep storage, used because we can't collapse SCC's in the
+    // predecessor graph by uniting the variables permanently, we can only do so
+    // for the successor graph.
+    std::vector<unsigned> VSSCCRep;
+    // Mapping from node to whether we have visited it during SCC finding yet.
+    std::vector<bool> Node2Visited;
+    // During variable substitution, we create unknowns to represent the unknown
+    // value that is a dereference of a variable.  These nodes are known as
+    // "ref" nodes (since they represent the value of dereferences).
+    unsigned FirstRefNode;
+    // During HVN, we create represent address taken nodes as if they were
+    // unknown (since HVN, unlike HU, does not evaluate unions).
+    unsigned FirstAdrNode;
+    // Current pointer equivalence class number
+    unsigned PEClass;
+    // Mapping from points-to sets to equivalence classes
+    typedef DenseMap<SparseBitVector<> *, unsigned, BitmapKeyInfo> BitVectorMap;
+    BitVectorMap Set2PEClass;
+    // Mapping from pointer equivalences to the representative node.  -1 if we
+    // have no representative node for this pointer equivalence class yet.
+    std::vector<int> PEClass2Node;
+    // Mapping from pointer equivalences to representative node.  This includes
+    // pointer equivalent but not location equivalent variables. -1 if we have
+    // no representative node for this pointer equivalence class yet.
+    std::vector<int> PENLEClass2Node;
+    // Union/Find for HCD
+    std::vector<unsigned> HCDSCCRep;
+    // HCD's offline-detected cycles; "Statically DeTected"
+    // -1 if not part of such a cycle, otherwise a representative node.
+    std::vector<int> SDT;
+    // Whether to use SDT (UniteNodes can use it during solving, but not before)
+    bool SDTActive;
+
+  public:
+    static char ID;
+    Andersens() : ModulePass(&ID) {}
+
+    bool runOnModule(Module &M) {
+      InitializeAliasAnalysis(this);
+      IdentifyObjects(M);
+      CollectConstraints(M);
+#undef DEBUG_TYPE
+#define DEBUG_TYPE "anders-aa-constraints"
+      DEBUG(PrintConstraints());
+#undef DEBUG_TYPE
+#define DEBUG_TYPE "anders-aa"
+      SolveConstraints();
+      DEBUG(PrintPointsToGraph());
+
+      // Free the constraints list, as we don't need it to respond to alias
+      // requests.
+      std::vector<Constraint>().swap(Constraints);
+      //These are needed for Print() (-analyze in opt)
+      //ObjectNodes.clear();
+      //ReturnNodes.clear();
+      //VarargNodes.clear();
+      return false;
+    }
+
+    void releaseMemory() {
+      // FIXME: Until we have transitively required passes working correctly,
+      // this cannot be enabled!  Otherwise, using -count-aa with the pass
+      // causes memory to be freed too early. :(
+#if 0
+      // The memory objects and ValueNodes data structures at the only ones that
+      // are still live after construction.
+      std::vector<Node>().swap(GraphNodes);
+      ValueNodes.clear();
+#endif
+    }
+
+    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
+      AliasAnalysis::getAnalysisUsage(AU);
+      AU.setPreservesAll();                         // Does not transform code
+    }
+
+    //------------------------------------------------
+    // Implement the AliasAnalysis API
+    //
+    AliasResult alias(const Value *V1, unsigned V1Size,
+                      const Value *V2, unsigned V2Size);
+    virtual ModRefResult getModRefInfo(CallSite CS, Value *P, unsigned Size);
+    virtual ModRefResult getModRefInfo(CallSite CS1, CallSite CS2);
+    void getMustAliases(Value *P, std::vector<Value*> &RetVals);
+    bool pointsToConstantMemory(const Value *P);
+
+    virtual void deleteValue(Value *V) {
+      ValueNodes.erase(V);
+      getAnalysis<AliasAnalysis>().deleteValue(V);
+    }
+
+    virtual void copyValue(Value *From, Value *To) {
+      ValueNodes[To] = ValueNodes[From];
+      getAnalysis<AliasAnalysis>().copyValue(From, To);
+    }
+
+  private:
+    /// getNode - Return the node corresponding to the specified pointer scalar.
+    ///
+    unsigned getNode(Value *V) {
+      if (Constant *C = dyn_cast<Constant>(V))
+        if (!isa<GlobalValue>(C))
+          return getNodeForConstantPointer(C);
+
+      DenseMap<Value*, unsigned>::iterator I = ValueNodes.find(V);
+      if (I == ValueNodes.end()) {
+#ifndef NDEBUG
+        V->dump();
+#endif
+        assert(0 && "Value does not have a node in the points-to graph!");
+      }
+      return I->second;
+    }
+
+    /// getObject - Return the node corresponding to the memory object for the
+    /// specified global or allocation instruction.
+    unsigned getObject(Value *V) const {
+      DenseMap<Value*, unsigned>::iterator I = ObjectNodes.find(V);
+      assert(I != ObjectNodes.end() &&
+             "Value does not have an object in the points-to graph!");
+      return I->second;
+    }
+
+    /// getReturnNode - Return the node representing the return value for the
+    /// specified function.
+    unsigned getReturnNode(Function *F) const {
+      DenseMap<Function*, unsigned>::iterator I = ReturnNodes.find(F);
+      assert(I != ReturnNodes.end() && "Function does not return a value!");
+      return I->second;
+    }
+
+    /// getVarargNode - Return the node representing the variable arguments
+    /// formal for the specified function.
+    unsigned getVarargNode(Function *F) const {
+      DenseMap<Function*, unsigned>::iterator I = VarargNodes.find(F);
+      assert(I != VarargNodes.end() && "Function does not take var args!");
+      return I->second;
+    }
+
+    /// getNodeValue - Get the node for the specified LLVM value and set the
+    /// value for it to be the specified value.
+    unsigned getNodeValue(Value &V) {
+      unsigned Index = getNode(&V);
+      GraphNodes[Index].setValue(&V);
+      return Index;
+    }
+
+    unsigned UniteNodes(unsigned First, unsigned Second,
+                        bool UnionByRank = true);
+    unsigned FindNode(unsigned Node);
+    unsigned FindNode(unsigned Node) const;
+
+    void IdentifyObjects(Module &M);
+    void CollectConstraints(Module &M);
+    bool AnalyzeUsesOfFunction(Value *);
+    void CreateConstraintGraph();
+    void OptimizeConstraints();
+    unsigned FindEquivalentNode(unsigned, unsigned);
+    void ClumpAddressTaken();
+    void RewriteConstraints();
+    void HU();
+    void HVN();
+    void HCD();
+    void Search(unsigned Node);
+    void UnitePointerEquivalences();
+    void SolveConstraints();
+    bool QueryNode(unsigned Node);
+    void Condense(unsigned Node);
+    void HUValNum(unsigned Node);
+    void HVNValNum(unsigned Node);
+    unsigned getNodeForConstantPointer(Constant *C);
+    unsigned getNodeForConstantPointerTarget(Constant *C);
+    void AddGlobalInitializerConstraints(unsigned, Constant *C);
+
+    void AddConstraintsForNonInternalLinkage(Function *F);
+    void AddConstraintsForCall(CallSite CS, Function *F);
+    bool AddConstraintsForExternalCall(CallSite CS, Function *F);
+
+
+    void PrintNode(const Node *N) const;
+    void PrintConstraints() const ;
+    void PrintConstraint(const Constraint &) const;
+    void PrintLabels() const;
+    void PrintPointsToGraph() const;
+
+    //===------------------------------------------------------------------===//
+    // Instruction visitation methods for adding constraints
+    //
+    friend class InstVisitor<Andersens>;
+    void visitReturnInst(ReturnInst &RI);
+    void visitInvokeInst(InvokeInst &II) { visitCallSite(CallSite(&II)); }
+    void visitCallInst(CallInst &CI) { visitCallSite(CallSite(&CI)); }
+    void visitCallSite(CallSite CS);
+    void visitAllocationInst(AllocationInst &AI);
+    void visitLoadInst(LoadInst &LI);
+    void visitStoreInst(StoreInst &SI);
+    void visitGetElementPtrInst(GetElementPtrInst &GEP);
+    void visitPHINode(PHINode &PN);
+    void visitCastInst(CastInst &CI);
+    void visitICmpInst(ICmpInst &ICI) {} // NOOP!
+    void visitFCmpInst(FCmpInst &ICI) {} // NOOP!
+    void visitSelectInst(SelectInst &SI);
+    void visitVAArg(VAArgInst &I);
+    void visitInstruction(Instruction &I);
+
+    //===------------------------------------------------------------------===//
+    // Implement Analyize interface
+    //
+    void print(std::ostream &O, const Module* M) const {
+      PrintPointsToGraph();
+    }
+  };
+}
+
+char Andersens::ID = 0;
+static RegisterPass<Andersens>
+X("anders-aa", "Andersen's Interprocedural Alias Analysis", false, true);
+static RegisterAnalysisGroup<AliasAnalysis> Y(X);
+
+// Initialize Timestamp Counter (static).
+unsigned Andersens::Node::Counter = 0;
+
+ModulePass *llvm::createAndersensPass() { return new Andersens(); }
+
+//===----------------------------------------------------------------------===//
+//                  AliasAnalysis Interface Implementation
+//===----------------------------------------------------------------------===//
+
+AliasAnalysis::AliasResult Andersens::alias(const Value *V1, unsigned V1Size,
+                                            const Value *V2, unsigned V2Size) {
+  Node *N1 = &GraphNodes[FindNode(getNode(const_cast<Value*>(V1)))];
+  Node *N2 = &GraphNodes[FindNode(getNode(const_cast<Value*>(V2)))];
+
+  // Check to see if the two pointers are known to not alias.  They don't alias
+  // if their points-to sets do not intersect.
+  if (!N1->intersectsIgnoring(N2, NullObject))
+    return NoAlias;
+
+  return AliasAnalysis::alias(V1, V1Size, V2, V2Size);
+}
+
+AliasAnalysis::ModRefResult
+Andersens::getModRefInfo(CallSite CS, Value *P, unsigned Size) {
+  // The only thing useful that we can contribute for mod/ref information is
+  // when calling external function calls: if we know that memory never escapes
+  // from the program, it cannot be modified by an external call.
+  //
+  // NOTE: This is not really safe, at least not when the entire program is not
+  // available.  The deal is that the external function could call back into the
+  // program and modify stuff.  We ignore this technical niggle for now.  This
+  // is, after all, a "research quality" implementation of Andersen's analysis.
+  if (Function *F = CS.getCalledFunction())
+    if (F->isDeclaration()) {
+      Node *N1 = &GraphNodes[FindNode(getNode(P))];
+
+      if (N1->PointsTo->empty())
+        return NoModRef;
+#if FULL_UNIVERSAL
+      if (!UniversalSet->PointsTo->test(FindNode(getNode(P))))
+        return NoModRef;  // Universal set does not contain P
+#else
+      if (!N1->PointsTo->test(UniversalSet))
+        return NoModRef;  // P doesn't point to the universal set.
+#endif
+    }
+
+  return AliasAnalysis::getModRefInfo(CS, P, Size);
+}
+
+AliasAnalysis::ModRefResult
+Andersens::getModRefInfo(CallSite CS1, CallSite CS2) {
+  return AliasAnalysis::getModRefInfo(CS1,CS2);
+}
+
+/// getMustAlias - We can provide must alias information if we know that a
+/// pointer can only point to a specific function or the null pointer.
+/// Unfortunately we cannot determine must-alias information for global
+/// variables or any other memory memory objects because we do not track whether
+/// a pointer points to the beginning of an object or a field of it.
+void Andersens::getMustAliases(Value *P, std::vector<Value*> &RetVals) {
+  Node *N = &GraphNodes[FindNode(getNode(P))];
+  if (N->PointsTo->count() == 1) {
+    Node *Pointee = &GraphNodes[N->PointsTo->find_first()];
+    // If a function is the only object in the points-to set, then it must be
+    // the destination.  Note that we can't handle global variables here,
+    // because we don't know if the pointer is actually pointing to a field of
+    // the global or to the beginning of it.
+    if (Value *V = Pointee->getValue()) {
+      if (Function *F = dyn_cast<Function>(V))
+        RetVals.push_back(F);
+    } else {
+      // If the object in the points-to set is the null object, then the null
+      // pointer is a must alias.
+      if (Pointee == &GraphNodes[NullObject])
+        RetVals.push_back(Constant::getNullValue(P->getType()));
+    }
+  }
+  AliasAnalysis::getMustAliases(P, RetVals);
+}
+
+/// pointsToConstantMemory - If we can determine that this pointer only points
+/// to constant memory, return true.  In practice, this means that if the
+/// pointer can only point to constant globals, functions, or the null pointer,
+/// return true.
+///
+bool Andersens::pointsToConstantMemory(const Value *P) {
+  Node *N = &GraphNodes[FindNode(getNode(const_cast<Value*>(P)))];
+  unsigned i;
+
+  for (SparseBitVector<>::iterator bi = N->PointsTo->begin();
+       bi != N->PointsTo->end();
+       ++bi) {
+    i = *bi;
+    Node *Pointee = &GraphNodes[i];
+    if (Value *V = Pointee->getValue()) {
+      if (!isa<GlobalValue>(V) || (isa<GlobalVariable>(V) &&
+                                   !cast<GlobalVariable>(V)->isConstant()))
+        return AliasAnalysis::pointsToConstantMemory(P);
+    } else {
+      if (i != NullObject)
+        return AliasAnalysis::pointsToConstantMemory(P);
+    }
+  }
+
+  return true;
+}
+
+//===----------------------------------------------------------------------===//
+//                       Object Identification Phase
+//===----------------------------------------------------------------------===//
+
+/// IdentifyObjects - This stage scans the program, adding an entry to the
+/// GraphNodes list for each memory object in the program (global stack or
+/// heap), and populates the ValueNodes and ObjectNodes maps for these objects.
+///
+void Andersens::IdentifyObjects(Module &M) {
+  unsigned NumObjects = 0;
+
+  // Object #0 is always the universal set: the object that we don't know
+  // anything about.
+  assert(NumObjects == UniversalSet && "Something changed!");
+  ++NumObjects;
+
+  // Object #1 always represents the null pointer.
+  assert(NumObjects == NullPtr && "Something changed!");
+  ++NumObjects;
+
+  // Object #2 always represents the null object (the object pointed to by null)
+  assert(NumObjects == NullObject && "Something changed!");
+  ++NumObjects;
+
+  // Add all the globals first.
+  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
+       I != E; ++I) {
+    ObjectNodes[I] = NumObjects++;
+    ValueNodes[I] = NumObjects++;
+  }
+
+  // Add nodes for all of the functions and the instructions inside of them.
+  for (Module::iterator F = M.begin(), E = M.end(); F != E; ++F) {
+    // The function itself is a memory object.
+    unsigned First = NumObjects;
+    ValueNodes[F] = NumObjects++;
+    if (isa<PointerType>(F->getFunctionType()->getReturnType()))
+      ReturnNodes[F] = NumObjects++;
+    if (F->getFunctionType()->isVarArg())
+      VarargNodes[F] = NumObjects++;
+
+
+    // Add nodes for all of the incoming pointer arguments.
+    for (Function::arg_iterator I = F->arg_begin(), E = F->arg_end();
+         I != E; ++I)
+      {
+        if (isa<PointerType>(I->getType()))
+          ValueNodes[I] = NumObjects++;
+      }
+    MaxK[First] = NumObjects - First;
+
+    // Scan the function body, creating a memory object for each heap/stack
+    // allocation in the body of the function and a node to represent all
+    // pointer values defined by instructions and used as operands.
+    for (inst_iterator II = inst_begin(F), E = inst_end(F); II != E; ++II) {
+      // If this is an heap or stack allocation, create a node for the memory
+      // object.
+      if (isa<PointerType>(II->getType())) {
+        ValueNodes[&*II] = NumObjects++;
+        if (AllocationInst *AI = dyn_cast<AllocationInst>(&*II))
+          ObjectNodes[AI] = NumObjects++;
+      }
+
+      // Calls to inline asm need to be added as well because the callee isn't
+      // referenced anywhere else.
+      if (CallInst *CI = dyn_cast<CallInst>(&*II)) {
+        Value *Callee = CI->getCalledValue();
+        if (isa<InlineAsm>(Callee))
+          ValueNodes[Callee] = NumObjects++;
+      }
+    }
+  }
+
+  // Now that we know how many objects to create, make them all now!
+  GraphNodes.resize(NumObjects);
+  NumNodes += NumObjects;
+}
+
+//===----------------------------------------------------------------------===//
+//                     Constraint Identification Phase
+//===----------------------------------------------------------------------===//
+
+/// getNodeForConstantPointer - Return the node corresponding to the constant
+/// pointer itself.
+unsigned Andersens::getNodeForConstantPointer(Constant *C) {
+  assert(isa<PointerType>(C->getType()) && "Not a constant pointer!");
+
+  if (isa<ConstantPointerNull>(C) || isa<UndefValue>(C))
+    return NullPtr;
+  else if (GlobalValue *GV = dyn_cast<GlobalValue>(C))
+    return getNode(GV);
+  else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(C)) {
+    switch (CE->getOpcode()) {
+    case Instruction::GetElementPtr:
+      return getNodeForConstantPointer(CE->getOperand(0));
+    case Instruction::IntToPtr:
+      return UniversalSet;
+    case Instruction::BitCast:
+      return getNodeForConstantPointer(CE->getOperand(0));
+    default:
+      cerr << "Constant Expr not yet handled: " << *CE << "\n";
+      assert(0);
+    }
+  } else {
+    assert(0 && "Unknown constant pointer!");
+  }
+  return 0;
+}
+
+/// getNodeForConstantPointerTarget - Return the node POINTED TO by the
+/// specified constant pointer.
+unsigned Andersens::getNodeForConstantPointerTarget(Constant *C) {
+  assert(isa<PointerType>(C->getType()) && "Not a constant pointer!");
+
+  if (isa<ConstantPointerNull>(C))
+    return NullObject;
+  else if (GlobalValue *GV = dyn_cast<GlobalValue>(C))
+    return getObject(GV);
+  else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(C)) {
+    switch (CE->getOpcode()) {
+    case Instruction::GetElementPtr:
+      return getNodeForConstantPointerTarget(CE->getOperand(0));
+    case Instruction::IntToPtr:
+      return UniversalSet;
+    case Instruction::BitCast:
+      return getNodeForConstantPointerTarget(CE->getOperand(0));
+    default:
+      cerr << "Constant Expr not yet handled: " << *CE << "\n";
+      assert(0);
+    }
+  } else {
+    assert(0 && "Unknown constant pointer!");
+  }
+  return 0;
+}
+
+/// AddGlobalInitializerConstraints - Add inclusion constraints for the memory
+/// object N, which contains values indicated by C.
+void Andersens::AddGlobalInitializerConstraints(unsigned NodeIndex,
+                                                Constant *C) {
+  if (C->getType()->isSingleValueType()) {
+    if (isa<PointerType>(C->getType()))
+      Constraints.push_back(Constraint(Constraint::Copy, NodeIndex,
+                                       getNodeForConstantPointer(C)));
+  } else if (C->isNullValue()) {
+    Constraints.push_back(Constraint(Constraint::Copy, NodeIndex,
+                                     NullObject));
+    return;
+  } else if (!isa<UndefValue>(C)) {
+    // If this is an array or struct, include constraints for each element.
+    assert(isa<ConstantArray>(C) || isa<ConstantStruct>(C));
+    for (unsigned i = 0, e = C->getNumOperands(); i != e; ++i)
+      AddGlobalInitializerConstraints(NodeIndex,
+                                      cast<Constant>(C->getOperand(i)));
+  }
+}
+
+/// AddConstraintsForNonInternalLinkage - If this function does not have
+/// internal linkage, realize that we can't trust anything passed into or
+/// returned by this function.
+void Andersens::AddConstraintsForNonInternalLinkage(Function *F) {
+  for (Function::arg_iterator I = F->arg_begin(), E = F->arg_end(); I != E; ++I)
+    if (isa<PointerType>(I->getType()))
+      // If this is an argument of an externally accessible function, the
+      // incoming pointer might point to anything.
+      Constraints.push_back(Constraint(Constraint::Copy, getNode(I),
+                                       UniversalSet));
+}
+
+/// AddConstraintsForCall - If this is a call to a "known" function, add the
+/// constraints and return true.  If this is a call to an unknown function,
+/// return false.
+bool Andersens::AddConstraintsForExternalCall(CallSite CS, Function *F) {
+  assert(F->isDeclaration() && "Not an external function!");
+
+  // These functions don't induce any points-to constraints.
+  if (F->getName() == "atoi" || F->getName() == "atof" ||
+      F->getName() == "atol" || F->getName() == "atoll" ||
+      F->getName() == "remove" || F->getName() == "unlink" ||
+      F->getName() == "rename" || F->getName() == "memcmp" ||
+      F->getName() == "llvm.memset" ||
+      F->getName() == "strcmp" || F->getName() == "strncmp" ||
+      F->getName() == "execl" || F->getName() == "execlp" ||
+      F->getName() == "execle" || F->getName() == "execv" ||
+      F->getName() == "execvp" || F->getName() == "chmod" ||
+      F->getName() == "puts" || F->getName() == "write" ||
+      F->getName() == "open" || F->getName() == "create" ||
+      F->getName() == "truncate" || F->getName() == "chdir" ||
+      F->getName() == "mkdir" || F->getName() == "rmdir" ||
+      F->getName() == "read" || F->getName() == "pipe" ||
+      F->getName() == "wait" || F->getName() == "time" ||
+      F->getName() == "stat" || F->getName() == "fstat" ||
+      F->getName() == "lstat" || F->getName() == "strtod" ||
+      F->getName() == "strtof" || F->getName() == "strtold" ||
+      F->getName() == "fopen" || F->getName() == "fdopen" ||
+      F->getName() == "freopen" ||
+      F->getName() == "fflush" || F->getName() == "feof" ||
+      F->getName() == "fileno" || F->getName() == "clearerr" ||
+      F->getName() == "rewind" || F->getName() == "ftell" ||
+      F->getName() == "ferror" || F->getName() == "fgetc" ||
+      F->getName() == "fgetc" || F->getName() == "_IO_getc" ||
+      F->getName() == "fwrite" || F->getName() == "fread" ||
+      F->getName() == "fgets" || F->getName() == "ungetc" ||
+      F->getName() == "fputc" ||
+      F->getName() == "fputs" || F->getName() == "putc" ||
+      F->getName() == "ftell" || F->getName() == "rewind" ||
+      F->getName() == "_IO_putc" || F->getName() == "fseek" ||
+      F->getName() == "fgetpos" || F->getName() == "fsetpos" ||
+      F->getName() == "printf" || F->getName() == "fprintf" ||
+      F->getName() == "sprintf" || F->getName() == "vprintf" ||
+      F->getName() == "vfprintf" || F->getName() == "vsprintf" ||
+      F->getName() == "scanf" || F->getName() == "fscanf" ||
+      F->getName() == "sscanf" || F->getName() == "__assert_fail" ||
+      F->getName() == "modf")
+    return true;
+
+
+  // These functions do induce points-to edges.
+  if (F->getName() == "llvm.memcpy" ||
+      F->getName() == "llvm.memmove" ||
+      F->getName() == "memmove") {
+
+    const FunctionType *FTy = F->getFunctionType();
+    if (FTy->getNumParams() > 1 && 
+        isa<PointerType>(FTy->getParamType(0)) &&
+        isa<PointerType>(FTy->getParamType(1))) {
+
+      // *Dest = *Src, which requires an artificial graph node to represent the
+      // constraint.  It is broken up into *Dest = temp, temp = *Src
+      unsigned FirstArg = getNode(CS.getArgument(0));
+      unsigned SecondArg = getNode(CS.getArgument(1));
+      unsigned TempArg = GraphNodes.size();
+      GraphNodes.push_back(Node());
+      Constraints.push_back(Constraint(Constraint::Store,
+                                       FirstArg, TempArg));
+      Constraints.push_back(Constraint(Constraint::Load,
+                                       TempArg, SecondArg));
+      // In addition, Dest = Src
+      Constraints.push_back(Constraint(Constraint::Copy,
+                                       FirstArg, SecondArg));
+      return true;
+    }
+  }
+
+  // Result = Arg0
+  if (F->getName() == "realloc" || F->getName() == "strchr" ||
+      F->getName() == "strrchr" || F->getName() == "strstr" ||
+      F->getName() == "strtok") {
+    const FunctionType *FTy = F->getFunctionType();
+    if (FTy->getNumParams() > 0 && 
+        isa<PointerType>(FTy->getParamType(0))) {
+      Constraints.push_back(Constraint(Constraint::Copy,
+                                       getNode(CS.getInstruction()),
+                                       getNode(CS.getArgument(0))));
+      return true;
+    }
+  }
+
+  return false;
+}
+
+
+
+/// AnalyzeUsesOfFunction - Look at all of the users of the specified function.
+/// If this is used by anything complex (i.e., the address escapes), return
+/// true.
+bool Andersens::AnalyzeUsesOfFunction(Value *V) {
+
+  if (!isa<PointerType>(V->getType())) return true;
+
+  for (Value::use_iterator UI = V->use_begin(), E = V->use_end(); UI != E; ++UI)
+    if (dyn_cast<LoadInst>(*UI)) {
+      return false;
+    } else if (StoreInst *SI = dyn_cast<StoreInst>(*UI)) {
+      if (V == SI->getOperand(1)) {
+        return false;
+      } else if (SI->getOperand(1)) {
+        return true;  // Storing the pointer
+      }
+    } else if (GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(*UI)) {
+      if (AnalyzeUsesOfFunction(GEP)) return true;
+    } else if (CallInst *CI = dyn_cast<CallInst>(*UI)) {
+      // Make sure that this is just the function being called, not that it is
+      // passing into the function.
+      for (unsigned i = 1, e = CI->getNumOperands(); i != e; ++i)
+        if (CI->getOperand(i) == V) return true;
+    } else if (InvokeInst *II = dyn_cast<InvokeInst>(*UI)) {
+      // Make sure that this is just the function being called, not that it is
+      // passing into the function.
+      for (unsigned i = 3, e = II->getNumOperands(); i != e; ++i)
+        if (II->getOperand(i) == V) return true;
+    } else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(*UI)) {
+      if (CE->getOpcode() == Instruction::GetElementPtr ||
+          CE->getOpcode() == Instruction::BitCast) {
+        if (AnalyzeUsesOfFunction(CE))
+          return true;
+      } else {
+        return true;
+      }
+    } else if (ICmpInst *ICI = dyn_cast<ICmpInst>(*UI)) {
+      if (!isa<ConstantPointerNull>(ICI->getOperand(1)))
+        return true;  // Allow comparison against null.
+    } else if (dyn_cast<FreeInst>(*UI)) {
+      return false;
+    } else {
+      return true;
+    }
+  return false;
+}
+
+/// CollectConstraints - This stage scans the program, adding a constraint to
+/// the Constraints list for each instruction in the program that induces a
+/// constraint, and setting up the initial points-to graph.
+///
+void Andersens::CollectConstraints(Module &M) {
+  // First, the universal set points to itself.
+  Constraints.push_back(Constraint(Constraint::AddressOf, UniversalSet,
+                                   UniversalSet));
+  Constraints.push_back(Constraint(Constraint::Store, UniversalSet,
+                                   UniversalSet));
+
+  // Next, the null pointer points to the null object.
+  Constraints.push_back(Constraint(Constraint::AddressOf, NullPtr, NullObject));
+
+  // Next, add any constraints on global variables and their initializers.
+  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
+       I != E; ++I) {
+    // Associate the address of the global object as pointing to the memory for
+    // the global: &G = <G memory>
+    unsigned ObjectIndex = getObject(I);
+    Node *Object = &GraphNodes[ObjectIndex];
+    Object->setValue(I);
+    Constraints.push_back(Constraint(Constraint::AddressOf, getNodeValue(*I),
+                                     ObjectIndex));
+
+    if (I->hasInitializer()) {
+      AddGlobalInitializerConstraints(ObjectIndex, I->getInitializer());
+    } else {
+      // If it doesn't have an initializer (i.e. it's defined in another
+      // translation unit), it points to the universal set.
+      Constraints.push_back(Constraint(Constraint::Copy, ObjectIndex,
+                                       UniversalSet));
+    }
+  }
+
+  for (Module::iterator F = M.begin(), E = M.end(); F != E; ++F) {
+    // Set up the return value node.
+    if (isa<PointerType>(F->getFunctionType()->getReturnType()))
+      GraphNodes[getReturnNode(F)].setValue(F);
+    if (F->getFunctionType()->isVarArg())
+      GraphNodes[getVarargNode(F)].setValue(F);
+
+    // Set up incoming argument nodes.
+    for (Function::arg_iterator I = F->arg_begin(), E = F->arg_end();
+         I != E; ++I)
+      if (isa<PointerType>(I->getType()))
+        getNodeValue(*I);
+
+    // At some point we should just add constraints for the escaping functions
+    // at solve time, but this slows down solving. For now, we simply mark
+    // address taken functions as escaping and treat them as external.
+    if (!F->hasLocalLinkage() || AnalyzeUsesOfFunction(F))
+      AddConstraintsForNonInternalLinkage(F);
+
+    if (!F->isDeclaration()) {
+      // Scan the function body, creating a memory object for each heap/stack
+      // allocation in the body of the function and a node to represent all
+      // pointer values defined by instructions and used as operands.
+      visit(F);
+    } else {
+      // External functions that return pointers return the universal set.
+      if (isa<PointerType>(F->getFunctionType()->getReturnType()))
+        Constraints.push_back(Constraint(Constraint::Copy,
+                                         getReturnNode(F),
+                                         UniversalSet));
+
+      // Any pointers that are passed into the function have the universal set
+      // stored into them.
+      for (Function::arg_iterator I = F->arg_begin(), E = F->arg_end();
+           I != E; ++I)
+        if (isa<PointerType>(I->getType())) {
+          // Pointers passed into external functions could have anything stored
+          // through them.
+          Constraints.push_back(Constraint(Constraint::Store, getNode(I),
+                                           UniversalSet));
+          // Memory objects passed into external function calls can have the
+          // universal set point to them.
+#if FULL_UNIVERSAL
+          Constraints.push_back(Constraint(Constraint::Copy,
+                                           UniversalSet,
+                                           getNode(I)));
+#else
+          Constraints.push_back(Constraint(Constraint::Copy,
+                                           getNode(I),
+                                           UniversalSet));
+#endif
+        }
+
+      // If this is an external varargs function, it can also store pointers
+      // into any pointers passed through the varargs section.
+      if (F->getFunctionType()->isVarArg())
+        Constraints.push_back(Constraint(Constraint::Store, getVarargNode(F),
+                                         UniversalSet));
+    }
+  }
+  NumConstraints += Constraints.size();
+}
+
+
+void Andersens::visitInstruction(Instruction &I) {
+#ifdef NDEBUG
+  return;          // This function is just a big assert.
+#endif
+  if (isa<BinaryOperator>(I))
+    return;
+  // Most instructions don't have any effect on pointer values.
+  switch (I.getOpcode()) {
+  case Instruction::Br:
+  case Instruction::Switch:
+  case Instruction::Unwind:
+  case Instruction::Unreachable:
+  case Instruction::Free:
+  case Instruction::ICmp:
+  case Instruction::FCmp:
+    return;
+  default:
+    // Is this something we aren't handling yet?
+    cerr << "Unknown instruction: " << I;
+    abort();
+  }
+}
+
+void Andersens::visitAllocationInst(AllocationInst &AI) {
+  unsigned ObjectIndex = getObject(&AI);
+  GraphNodes[ObjectIndex].setValue(&AI);
+  Constraints.push_back(Constraint(Constraint::AddressOf, getNodeValue(AI),
+                                   ObjectIndex));
+}
+
+void Andersens::visitReturnInst(ReturnInst &RI) {
+  if (RI.getNumOperands() && isa<PointerType>(RI.getOperand(0)->getType()))
+    // return V   -->   <Copy/retval{F}/v>
+    Constraints.push_back(Constraint(Constraint::Copy,
+                                     getReturnNode(RI.getParent()->getParent()),
+                                     getNode(RI.getOperand(0))));
+}
+
+void Andersens::visitLoadInst(LoadInst &LI) {
+  if (isa<PointerType>(LI.getType()))
+    // P1 = load P2  -->  <Load/P1/P2>
+    Constraints.push_back(Constraint(Constraint::Load, getNodeValue(LI),
+                                     getNode(LI.getOperand(0))));
+}
+
+void Andersens::visitStoreInst(StoreInst &SI) {
+  if (isa<PointerType>(SI.getOperand(0)->getType()))
+    // store P1, P2  -->  <Store/P2/P1>
+    Constraints.push_back(Constraint(Constraint::Store,
+                                     getNode(SI.getOperand(1)),
+                                     getNode(SI.getOperand(0))));
+}
+
+void Andersens::visitGetElementPtrInst(GetElementPtrInst &GEP) {
+  // P1 = getelementptr P2, ... --> <Copy/P1/P2>
+  Constraints.push_back(Constraint(Constraint::Copy, getNodeValue(GEP),
+                                   getNode(GEP.getOperand(0))));
+}
+
+void Andersens::visitPHINode(PHINode &PN) {
+  if (isa<PointerType>(PN.getType())) {
+    unsigned PNN = getNodeValue(PN);
+    for (unsigned i = 0, e = PN.getNumIncomingValues(); i != e; ++i)
+      // P1 = phi P2, P3  -->  <Copy/P1/P2>, <Copy/P1/P3>, ...
+      Constraints.push_back(Constraint(Constraint::Copy, PNN,
+                                       getNode(PN.getIncomingValue(i))));
+  }
+}
+
+void Andersens::visitCastInst(CastInst &CI) {
+  Value *Op = CI.getOperand(0);
+  if (isa<PointerType>(CI.getType())) {
+    if (isa<PointerType>(Op->getType())) {
+      // P1 = cast P2  --> <Copy/P1/P2>
+      Constraints.push_back(Constraint(Constraint::Copy, getNodeValue(CI),
+                                       getNode(CI.getOperand(0))));
+    } else {
+      // P1 = cast int --> <Copy/P1/Univ>
+#if 0
+      Constraints.push_back(Constraint(Constraint::Copy, getNodeValue(CI),
+                                       UniversalSet));
+#else
+      getNodeValue(CI);
+#endif
+    }
+  } else if (isa<PointerType>(Op->getType())) {
+    // int = cast P1 --> <Copy/Univ/P1>
+#if 0
+    Constraints.push_back(Constraint(Constraint::Copy,
+                                     UniversalSet,
+                                     getNode(CI.getOperand(0))));
+#else
+    getNode(CI.getOperand(0));
+#endif
+  }
+}
+
+void Andersens::visitSelectInst(SelectInst &SI) {
+  if (isa<PointerType>(SI.getType())) {
+    unsigned SIN = getNodeValue(SI);
+    // P1 = select C, P2, P3   ---> <Copy/P1/P2>, <Copy/P1/P3>
+    Constraints.push_back(Constraint(Constraint::Copy, SIN,
+                                     getNode(SI.getOperand(1))));
+    Constraints.push_back(Constraint(Constraint::Copy, SIN,
+                                     getNode(SI.getOperand(2))));
+  }
+}
+
+void Andersens::visitVAArg(VAArgInst &I) {
+  assert(0 && "vaarg not handled yet!");
+}
+
+/// AddConstraintsForCall - Add constraints for a call with actual arguments
+/// specified by CS to the function specified by F.  Note that the types of
+/// arguments might not match up in the case where this is an indirect call and
+/// the function pointer has been casted.  If this is the case, do something
+/// reasonable.
+void Andersens::AddConstraintsForCall(CallSite CS, Function *F) {
+  Value *CallValue = CS.getCalledValue();
+  bool IsDeref = F == NULL;
+
+  // If this is a call to an external function, try to handle it directly to get
+  // some taste of context sensitivity.
+  if (F && F->isDeclaration() && AddConstraintsForExternalCall(CS, F))
+    return;
+
+  if (isa<PointerType>(CS.getType())) {
+    unsigned CSN = getNode(CS.getInstruction());
+    if (!F || isa<PointerType>(F->getFunctionType()->getReturnType())) {
+      if (IsDeref)
+        Constraints.push_back(Constraint(Constraint::Load, CSN,
+                                         getNode(CallValue), CallReturnPos));
+      else
+        Constraints.push_back(Constraint(Constraint::Copy, CSN,
+                                         getNode(CallValue) + CallReturnPos));
+    } else {
+      // If the function returns a non-pointer value, handle this just like we
+      // treat a nonpointer cast to pointer.
+      Constraints.push_back(Constraint(Constraint::Copy, CSN,
+                                       UniversalSet));
+    }
+  } else if (F && isa<PointerType>(F->getFunctionType()->getReturnType())) {
+#if FULL_UNIVERSAL
+    Constraints.push_back(Constraint(Constraint::Copy,
+                                     UniversalSet,
+                                     getNode(CallValue) + CallReturnPos));
+#else
+    Constraints.push_back(Constraint(Constraint::Copy,
+                                      getNode(CallValue) + CallReturnPos,
+                                      UniversalSet));
+#endif
+                          
+    
+  }
+
+  CallSite::arg_iterator ArgI = CS.arg_begin(), ArgE = CS.arg_end();
+  bool external = !F ||  F->isDeclaration();
+  if (F) {
+    // Direct Call
+    Function::arg_iterator AI = F->arg_begin(), AE = F->arg_end();
+    for (; AI != AE && ArgI != ArgE; ++AI, ++ArgI) 
+      {
+#if !FULL_UNIVERSAL
+        if (external && isa<PointerType>((*ArgI)->getType())) 
+          {
+            // Add constraint that ArgI can now point to anything due to
+            // escaping, as can everything it points to. The second portion of
+            // this should be taken care of by universal = *universal
+            Constraints.push_back(Constraint(Constraint::Copy,
+                                             getNode(*ArgI),
+                                             UniversalSet));
+          }
+#endif
+        if (isa<PointerType>(AI->getType())) {
+          if (isa<PointerType>((*ArgI)->getType())) {
+            // Copy the actual argument into the formal argument.
+            Constraints.push_back(Constraint(Constraint::Copy, getNode(AI),
+                                             getNode(*ArgI)));
+          } else {
+            Constraints.push_back(Constraint(Constraint::Copy, getNode(AI),
+                                             UniversalSet));
+          }
+        } else if (isa<PointerType>((*ArgI)->getType())) {
+#if FULL_UNIVERSAL
+          Constraints.push_back(Constraint(Constraint::Copy,
+                                           UniversalSet,
+                                           getNode(*ArgI)));
+#else
+          Constraints.push_back(Constraint(Constraint::Copy,
+                                           getNode(*ArgI),
+                                           UniversalSet));
+#endif
+        }
+      }
+  } else {
+    //Indirect Call
+    unsigned ArgPos = CallFirstArgPos;
+    for (; ArgI != ArgE; ++ArgI) {
+      if (isa<PointerType>((*ArgI)->getType())) {
+        // Copy the actual argument into the formal argument.
+        Constraints.push_back(Constraint(Constraint::Store,
+                                         getNode(CallValue),
+                                         getNode(*ArgI), ArgPos++));
+      } else {
+        Constraints.push_back(Constraint(Constraint::Store,
+                                         getNode (CallValue),
+                                         UniversalSet, ArgPos++));
+      }
+    }
+  }
+  // Copy all pointers passed through the varargs section to the varargs node.
+  if (F && F->getFunctionType()->isVarArg())
+    for (; ArgI != ArgE; ++ArgI)
+      if (isa<PointerType>((*ArgI)->getType()))
+        Constraints.push_back(Constraint(Constraint::Copy, getVarargNode(F),
+                                         getNode(*ArgI)));
+  // If more arguments are passed in than we track, just drop them on the floor.
+}
+
+void Andersens::visitCallSite(CallSite CS) {
+  if (isa<PointerType>(CS.getType()))
+    getNodeValue(*CS.getInstruction());
+
+  if (Function *F = CS.getCalledFunction()) {
+    AddConstraintsForCall(CS, F);
+  } else {
+    AddConstraintsForCall(CS, NULL);
+  }
+}
+
+//===----------------------------------------------------------------------===//
+//                         Constraint Solving Phase
+//===----------------------------------------------------------------------===//
+
+/// intersects - Return true if the points-to set of this node intersects
+/// with the points-to set of the specified node.
+bool Andersens::Node::intersects(Node *N) const {
+  return PointsTo->intersects(N->PointsTo);
+}
+
+/// intersectsIgnoring - Return true if the points-to set of this node
+/// intersects with the points-to set of the specified node on any nodes
+/// except for the specified node to ignore.
+bool Andersens::Node::intersectsIgnoring(Node *N, unsigned Ignoring) const {
+  // TODO: If we are only going to call this with the same value for Ignoring,
+  // we should move the special values out of the points-to bitmap.
+  bool WeHadIt = PointsTo->test(Ignoring);
+  bool NHadIt = N->PointsTo->test(Ignoring);
+  bool Result = false;
+  if (WeHadIt)
+    PointsTo->reset(Ignoring);
+  if (NHadIt)
+    N->PointsTo->reset(Ignoring);
+  Result = PointsTo->intersects(N->PointsTo);
+  if (WeHadIt)
+    PointsTo->set(Ignoring);
+  if (NHadIt)
+    N->PointsTo->set(Ignoring);
+  return Result;
+}
+
+void dumpToDOUT(SparseBitVector<> *bitmap) {
+#ifndef NDEBUG
+  dump(*bitmap, DOUT);
+#endif
+}
+
+
+/// Clump together address taken variables so that the points-to sets use up
+/// less space and can be operated on faster.
+
+void Andersens::ClumpAddressTaken() {
+#undef DEBUG_TYPE
+#define DEBUG_TYPE "anders-aa-renumber"
+  std::vector<unsigned> Translate;
+  std::vector<Node> NewGraphNodes;
+
+  Translate.resize(GraphNodes.size());
+  unsigned NewPos = 0;
+
+  for (unsigned i = 0; i < Constraints.size(); ++i) {
+    Constraint &C = Constraints[i];
+    if (C.Type == Constraint::AddressOf) {
+      GraphNodes[C.Src].AddressTaken = true;
+    }
+  }
+  for (unsigned i = 0; i < NumberSpecialNodes; ++i) {
+    unsigned Pos = NewPos++;
+    Translate[i] = Pos;
+    NewGraphNodes.push_back(GraphNodes[i]);
+    DOUT << "Renumbering node " << i << " to node " << Pos << "\n";
+  }
+
+  // I believe this ends up being faster than making two vectors and splicing
+  // them.
+  for (unsigned i = NumberSpecialNodes; i < GraphNodes.size(); ++i) {
+    if (GraphNodes[i].AddressTaken) {
+      unsigned Pos = NewPos++;
+      Translate[i] = Pos;
+      NewGraphNodes.push_back(GraphNodes[i]);
+      DOUT << "Renumbering node " << i << " to node " << Pos << "\n";
+    }
+  }
+
+  for (unsigned i = NumberSpecialNodes; i < GraphNodes.size(); ++i) {
+    if (!GraphNodes[i].AddressTaken) {
+      unsigned Pos = NewPos++;
+      Translate[i] = Pos;
+      NewGraphNodes.push_back(GraphNodes[i]);
+      DOUT << "Renumbering node " << i << " to node " << Pos << "\n";
+    }
+  }
+
+  for (DenseMap<Value*, unsigned>::iterator Iter = ValueNodes.begin();
+       Iter != ValueNodes.end();
+       ++Iter)
+    Iter->second = Translate[Iter->second];
+
+  for (DenseMap<Value*, unsigned>::iterator Iter = ObjectNodes.begin();
+       Iter != ObjectNodes.end();
+       ++Iter)
+    Iter->second = Translate[Iter->second];
+
+  for (DenseMap<Function*, unsigned>::iterator Iter = ReturnNodes.begin();
+       Iter != ReturnNodes.end();
+       ++Iter)
+    Iter->second = Translate[Iter->second];
+
+  for (DenseMap<Function*, unsigned>::iterator Iter = VarargNodes.begin();
+       Iter != VarargNodes.end();
+       ++Iter)
+    Iter->second = Translate[Iter->second];
+
+  for (unsigned i = 0; i < Constraints.size(); ++i) {
+    Constraint &C = Constraints[i];
+    C.Src = Translate[C.Src];
+    C.Dest = Translate[C.Dest];
+  }
+
+  GraphNodes.swap(NewGraphNodes);
+#undef DEBUG_TYPE
+#define DEBUG_TYPE "anders-aa"
+}
+
+/// The technique used here is described in "Exploiting Pointer and Location
+/// Equivalence to Optimize Pointer Analysis. In the 14th International Static
+/// Analysis Symposium (SAS), August 2007."  It is known as the "HVN" algorithm,
+/// and is equivalent to value numbering the collapsed constraint graph without
+/// evaluating unions.  This is used as a pre-pass to HU in order to resolve
+/// first order pointer dereferences and speed up/reduce memory usage of HU.
+/// Running both is equivalent to HRU without the iteration
+/// HVN in more detail:
+/// Imagine the set of constraints was simply straight line code with no loops
+/// (we eliminate cycles, so there are no loops), such as:
+/// E = &D
+/// E = &C
+/// E = F
+/// F = G
+/// G = F
+/// Applying value numbering to this code tells us:
+/// G == F == E
+///
+/// For HVN, this is as far as it goes.  We assign new value numbers to every
+/// "address node", and every "reference node".
+/// To get the optimal result for this, we use a DFS + SCC (since all nodes in a
+/// cycle must have the same value number since the = operation is really
+/// inclusion, not overwrite), and value number nodes we receive points-to sets
+/// before we value our own node.
+/// The advantage of HU over HVN is that HU considers the inclusion property, so
+/// that if you have
+/// E = &D
+/// E = &C
+/// E = F
+/// F = G
+/// F = &D
+/// G = F
+/// HU will determine that G == F == E.  HVN will not, because it cannot prove
+/// that the points to information ends up being the same because they all
+/// receive &D from E anyway.
+
+void Andersens::HVN() {
+  DOUT << "Beginning HVN\n";
+  // Build a predecessor graph.  This is like our constraint graph with the
+  // edges going in the opposite direction, and there are edges for all the
+  // constraints, instead of just copy constraints.  We also build implicit
+  // edges for constraints are implied but not explicit.  I.E for the constraint
+  // a = &b, we add implicit edges *a = b.  This helps us capture more cycles
+  for (unsigned i = 0, e = Constraints.size(); i != e; ++i) {
+    Constraint &C = Constraints[i];
+    if (C.Type == Constraint::AddressOf) {
+      GraphNodes[C.Src].AddressTaken = true;
+      GraphNodes[C.Src].Direct = false;
+
+      // Dest = &src edge
+      unsigned AdrNode = C.Src + FirstAdrNode;
+      if (!GraphNodes[C.Dest].PredEdges)
+        GraphNodes[C.Dest].PredEdges = new SparseBitVector<>;
+      GraphNodes[C.Dest].PredEdges->set(AdrNode);
+
+      // *Dest = src edge
+      unsigned RefNode = C.Dest + FirstRefNode;
+      if (!GraphNodes[RefNode].ImplicitPredEdges)
+        GraphNodes[RefNode].ImplicitPredEdges = new SparseBitVector<>;
+      GraphNodes[RefNode].ImplicitPredEdges->set(C.Src);
+    } else if (C.Type == Constraint::Load) {
+      if (C.Offset == 0) {
+        // dest = *src edge
+        if (!GraphNodes[C.Dest].PredEdges)
+          GraphNodes[C.Dest].PredEdges = new SparseBitVector<>;
+        GraphNodes[C.Dest].PredEdges->set(C.Src + FirstRefNode);
+      } else {
+        GraphNodes[C.Dest].Direct = false;
+      }
+    } else if (C.Type == Constraint::Store) {
+      if (C.Offset == 0) {
+        // *dest = src edge
+        unsigned RefNode = C.Dest + FirstRefNode;
+        if (!GraphNodes[RefNode].PredEdges)
+          GraphNodes[RefNode].PredEdges = new SparseBitVector<>;
+        GraphNodes[RefNode].PredEdges->set(C.Src);
+      }
+    } else {
+      // Dest = Src edge and *Dest = *Src edge
+      if (!GraphNodes[C.Dest].PredEdges)
+        GraphNodes[C.Dest].PredEdges = new SparseBitVector<>;
+      GraphNodes[C.Dest].PredEdges->set(C.Src);
+      unsigned RefNode = C.Dest + FirstRefNode;
+      if (!GraphNodes[RefNode].ImplicitPredEdges)
+        GraphNodes[RefNode].ImplicitPredEdges = new SparseBitVector<>;
+      GraphNodes[RefNode].ImplicitPredEdges->set(C.Src + FirstRefNode);
+    }
+  }
+  PEClass = 1;
+  // Do SCC finding first to condense our predecessor graph
+  DFSNumber = 0;
+  Node2DFS.insert(Node2DFS.begin(), GraphNodes.size(), 0);
+  Node2Deleted.insert(Node2Deleted.begin(), GraphNodes.size(), false);
+  Node2Visited.insert(Node2Visited.begin(), GraphNodes.size(), false);
+
+  for (unsigned i = 0; i < FirstRefNode; ++i) {
+    unsigned Node = VSSCCRep[i];
+    if (!Node2Visited[Node])
+      HVNValNum(Node);
+  }
+  for (BitVectorMap::iterator Iter = Set2PEClass.begin();
+       Iter != Set2PEClass.end();
+       ++Iter)
+    delete Iter->first;
+  Set2PEClass.clear();
+  Node2DFS.clear();
+  Node2Deleted.clear();
+  Node2Visited.clear();
+  DOUT << "Finished HVN\n";
+
+}
+
+/// This is the workhorse of HVN value numbering. We combine SCC finding at the
+/// same time because it's easy.
+void Andersens::HVNValNum(unsigned NodeIndex) {
+  unsigned MyDFS = DFSNumber++;
+  Node *N = &GraphNodes[NodeIndex];
+  Node2Visited[NodeIndex] = true;
+  Node2DFS[NodeIndex] = MyDFS;
+
+  // First process all our explicit edges
+  if (N->PredEdges)
+    for (SparseBitVector<>::iterator Iter = N->PredEdges->begin();
+         Iter != N->PredEdges->end();
+         ++Iter) {
+      unsigned j = VSSCCRep[*Iter];
+      if (!Node2Deleted[j]) {
+        if (!Node2Visited[j])
+          HVNValNum(j);
+        if (Node2DFS[NodeIndex] > Node2DFS[j])
+          Node2DFS[NodeIndex] = Node2DFS[j];
+      }
+    }
+
+  // Now process all the implicit edges
+  if (N->ImplicitPredEdges)
+    for (SparseBitVector<>::iterator Iter = N->ImplicitPredEdges->begin();
+         Iter != N->ImplicitPredEdges->end();
+         ++Iter) {
+      unsigned j = VSSCCRep[*Iter];
+      if (!Node2Deleted[j]) {
+        if (!Node2Visited[j])
+          HVNValNum(j);
+        if (Node2DFS[NodeIndex] > Node2DFS[j])
+          Node2DFS[NodeIndex] = Node2DFS[j];
+      }
+    }
+
+  // See if we found any cycles
+  if (MyDFS == Node2DFS[NodeIndex]) {
+    while (!SCCStack.empty() && Node2DFS[SCCStack.top()] >= MyDFS) {
+      unsigned CycleNodeIndex = SCCStack.top();
+      Node *CycleNode = &GraphNodes[CycleNodeIndex];
+      VSSCCRep[CycleNodeIndex] = NodeIndex;
+      // Unify the nodes
+      N->Direct &= CycleNode->Direct;
+
+      if (CycleNode->PredEdges) {
+        if (!N->PredEdges)
+          N->PredEdges = new SparseBitVector<>;
+        *(N->PredEdges) |= CycleNode->PredEdges;
+        delete CycleNode->PredEdges;
+        CycleNode->PredEdges = NULL;
+      }
+      if (CycleNode->ImplicitPredEdges) {
+        if (!N->ImplicitPredEdges)
+          N->ImplicitPredEdges = new SparseBitVector<>;
+        *(N->ImplicitPredEdges) |= CycleNode->ImplicitPredEdges;
+        delete CycleNode->ImplicitPredEdges;
+        CycleNode->ImplicitPredEdges = NULL;
+      }
+
+      SCCStack.pop();
+    }
+
+    Node2Deleted[NodeIndex] = true;
+
+    if (!N->Direct) {
+      GraphNodes[NodeIndex].PointerEquivLabel = PEClass++;
+      return;
+    }
+
+    // Collect labels of successor nodes
+    bool AllSame = true;
+    unsigned First = ~0;
+    SparseBitVector<> *Labels = new SparseBitVector<>;
+    bool Used = false;
+
+    if (N->PredEdges)
+      for (SparseBitVector<>::iterator Iter = N->PredEdges->begin();
+           Iter != N->PredEdges->end();
+         ++Iter) {
+        unsigned j = VSSCCRep[*Iter];
+        unsigned Label = GraphNodes[j].PointerEquivLabel;
+        // Ignore labels that are equal to us or non-pointers
+        if (j == NodeIndex || Label == 0)
+          continue;
+        if (First == (unsigned)~0)
+          First = Label;
+        else if (First != Label)
+          AllSame = false;
+        Labels->set(Label);
+    }
+
+    // We either have a non-pointer, a copy of an existing node, or a new node.
+    // Assign the appropriate pointer equivalence label.
+    if (Labels->empty()) {
+      GraphNodes[NodeIndex].PointerEquivLabel = 0;
+    } else if (AllSame) {
+      GraphNodes[NodeIndex].PointerEquivLabel = First;
+    } else {
+      GraphNodes[NodeIndex].PointerEquivLabel = Set2PEClass[Labels];
+      if (GraphNodes[NodeIndex].PointerEquivLabel == 0) {
+        unsigned EquivClass = PEClass++;
+        Set2PEClass[Labels] = EquivClass;
+        GraphNodes[NodeIndex].PointerEquivLabel = EquivClass;
+        Used = true;
+      }
+    }
+    if (!Used)
+      delete Labels;
+  } else {
+    SCCStack.push(NodeIndex);
+  }
+}
+
+/// The technique used here is described in "Exploiting Pointer and Location
+/// Equivalence to Optimize Pointer Analysis. In the 14th International Static
+/// Analysis Symposium (SAS), August 2007."  It is known as the "HU" algorithm,
+/// and is equivalent to value numbering the collapsed constraint graph
+/// including evaluating unions.
+void Andersens::HU() {
+  DOUT << "Beginning HU\n";
+  // Build a predecessor graph.  This is like our constraint graph with the
+  // edges going in the opposite direction, and there are edges for all the
+  // constraints, instead of just copy constraints.  We also build implicit
+  // edges for constraints are implied but not explicit.  I.E for the constraint
+  // a = &b, we add implicit edges *a = b.  This helps us capture more cycles
+  for (unsigned i = 0, e = Constraints.size(); i != e; ++i) {
+    Constraint &C = Constraints[i];
+    if (C.Type == Constraint::AddressOf) {
+      GraphNodes[C.Src].AddressTaken = true;
+      GraphNodes[C.Src].Direct = false;
+
+      GraphNodes[C.Dest].PointsTo->set(C.Src);
+      // *Dest = src edge
+      unsigned RefNode = C.Dest + FirstRefNode;
+      if (!GraphNodes[RefNode].ImplicitPredEdges)
+        GraphNodes[RefNode].ImplicitPredEdges = new SparseBitVector<>;
+      GraphNodes[RefNode].ImplicitPredEdges->set(C.Src);
+      GraphNodes[C.Src].PointedToBy->set(C.Dest);
+    } else if (C.Type == Constraint::Load) {
+      if (C.Offset == 0) {
+        // dest = *src edge
+        if (!GraphNodes[C.Dest].PredEdges)
+          GraphNodes[C.Dest].PredEdges = new SparseBitVector<>;
+        GraphNodes[C.Dest].PredEdges->set(C.Src + FirstRefNode);
+      } else {
+        GraphNodes[C.Dest].Direct = false;
+      }
+    } else if (C.Type == Constraint::Store) {
+      if (C.Offset == 0) {
+        // *dest = src edge
+        unsigned RefNode = C.Dest + FirstRefNode;
+        if (!GraphNodes[RefNode].PredEdges)
+          GraphNodes[RefNode].PredEdges = new SparseBitVector<>;
+        GraphNodes[RefNode].PredEdges->set(C.Src);
+      }
+    } else {
+      // Dest = Src edge and *Dest = *Src edg
+      if (!GraphNodes[C.Dest].PredEdges)
+        GraphNodes[C.Dest].PredEdges = new SparseBitVector<>;
+      GraphNodes[C.Dest].PredEdges->set(C.Src);
+      unsigned RefNode = C.Dest + FirstRefNode;
+      if (!GraphNodes[RefNode].ImplicitPredEdges)
+        GraphNodes[RefNode].ImplicitPredEdges = new SparseBitVector<>;
+      GraphNodes[RefNode].ImplicitPredEdges->set(C.Src + FirstRefNode);
+    }
+  }
+  PEClass = 1;
+  // Do SCC finding first to condense our predecessor graph
+  DFSNumber = 0;
+  Node2DFS.insert(Node2DFS.begin(), GraphNodes.size(), 0);
+  Node2Deleted.insert(Node2Deleted.begin(), GraphNodes.size(), false);
+  Node2Visited.insert(Node2Visited.begin(), GraphNodes.size(), false);
+
+  for (unsigned i = 0; i < FirstRefNode; ++i) {
+    if (FindNode(i) == i) {
+      unsigned Node = VSSCCRep[i];
+      if (!Node2Visited[Node])
+        Condense(Node);
+    }
+  }
+
+  // Reset tables for actual labeling
+  Node2DFS.clear();
+  Node2Visited.clear();
+  Node2Deleted.clear();
+  // Pre-grow our densemap so that we don't get really bad behavior
+  Set2PEClass.resize(GraphNodes.size());
+
+  // Visit the condensed graph and generate pointer equivalence labels.
+  Node2Visited.insert(Node2Visited.begin(), GraphNodes.size(), false);
+  for (unsigned i = 0; i < FirstRefNode; ++i) {
+    if (FindNode(i) == i) {
+      unsigned Node = VSSCCRep[i];
+      if (!Node2Visited[Node])
+        HUValNum(Node);
+    }
+  }
+  // PEClass nodes will be deleted by the deleting of N->PointsTo in our caller.
+  Set2PEClass.clear();
+  DOUT << "Finished HU\n";
+}
+
+
+/// Implementation of standard Tarjan SCC algorithm as modified by Nuutilla.
+void Andersens::Condense(unsigned NodeIndex) {
+  unsigned MyDFS = DFSNumber++;
+  Node *N = &GraphNodes[NodeIndex];
+  Node2Visited[NodeIndex] = true;
+  Node2DFS[NodeIndex] = MyDFS;
+
+  // First process all our explicit edges
+  if (N->PredEdges)
+    for (SparseBitVector<>::iterator Iter = N->PredEdges->begin();
+         Iter != N->PredEdges->end();
+         ++Iter) {
+      unsigned j = VSSCCRep[*Iter];
+      if (!Node2Deleted[j]) {
+        if (!Node2Visited[j])
+          Condense(j);
+        if (Node2DFS[NodeIndex] > Node2DFS[j])
+          Node2DFS[NodeIndex] = Node2DFS[j];
+      }
+    }
+
+  // Now process all the implicit edges
+  if (N->ImplicitPredEdges)
+    for (SparseBitVector<>::iterator Iter = N->ImplicitPredEdges->begin();
+         Iter != N->ImplicitPredEdges->end();
+         ++Iter) {
+      unsigned j = VSSCCRep[*Iter];
+      if (!Node2Deleted[j]) {
+        if (!Node2Visited[j])
+          Condense(j);
+        if (Node2DFS[NodeIndex] > Node2DFS[j])
+          Node2DFS[NodeIndex] = Node2DFS[j];
+      }
+    }
+
+  // See if we found any cycles
+  if (MyDFS == Node2DFS[NodeIndex]) {
+    while (!SCCStack.empty() && Node2DFS[SCCStack.top()] >= MyDFS) {
+      unsigned CycleNodeIndex = SCCStack.top();
+      Node *CycleNode = &GraphNodes[CycleNodeIndex];
+      VSSCCRep[CycleNodeIndex] = NodeIndex;
+      // Unify the nodes
+      N->Direct &= CycleNode->Direct;
+
+      *(N->PointsTo) |= CycleNode->PointsTo;
+      delete CycleNode->PointsTo;
+      CycleNode->PointsTo = NULL;
+      if (CycleNode->PredEdges) {
+        if (!N->PredEdges)
+          N->PredEdges = new SparseBitVector<>;
+        *(N->PredEdges) |= CycleNode->PredEdges;
+        delete CycleNode->PredEdges;
+        CycleNode->PredEdges = NULL;
+      }
+      if (CycleNode->ImplicitPredEdges) {
+        if (!N->ImplicitPredEdges)
+          N->ImplicitPredEdges = new SparseBitVector<>;
+        *(N->ImplicitPredEdges) |= CycleNode->ImplicitPredEdges;
+        delete CycleNode->ImplicitPredEdges;
+        CycleNode->ImplicitPredEdges = NULL;
+      }
+      SCCStack.pop();
+    }
+
+    Node2Deleted[NodeIndex] = true;
+
+    // Set up number of incoming edges for other nodes
+    if (N->PredEdges)
+      for (SparseBitVector<>::iterator Iter = N->PredEdges->begin();
+           Iter != N->PredEdges->end();
+           ++Iter)
+        ++GraphNodes[VSSCCRep[*Iter]].NumInEdges;
+  } else {
+    SCCStack.push(NodeIndex);
+  }
+}
+
+void Andersens::HUValNum(unsigned NodeIndex) {
+  Node *N = &GraphNodes[NodeIndex];
+  Node2Visited[NodeIndex] = true;
+
+  // Eliminate dereferences of non-pointers for those non-pointers we have
+  // already identified.  These are ref nodes whose non-ref node:
+  // 1. Has already been visited determined to point to nothing (and thus, a
+  // dereference of it must point to nothing)
+  // 2. Any direct node with no predecessor edges in our graph and with no
+  // points-to set (since it can't point to anything either, being that it
+  // receives no points-to sets and has none).
+  if (NodeIndex >= FirstRefNode) {
+    unsigned j = VSSCCRep[FindNode(NodeIndex - FirstRefNode)];
+    if ((Node2Visited[j] && !GraphNodes[j].PointerEquivLabel)
+        || (GraphNodes[j].Direct && !GraphNodes[j].PredEdges
+            && GraphNodes[j].PointsTo->empty())){
+      return;
+    }
+  }
+    // Process all our explicit edges
+  if (N->PredEdges)
+    for (SparseBitVector<>::iterator Iter = N->PredEdges->begin();
+         Iter != N->PredEdges->end();
+         ++Iter) {
+      unsigned j = VSSCCRep[*Iter];
+      if (!Node2Visited[j])
+        HUValNum(j);
+
+      // If this edge turned out to be the same as us, or got no pointer
+      // equivalence label (and thus points to nothing) , just decrement our
+      // incoming edges and continue.
+      if (j == NodeIndex || GraphNodes[j].PointerEquivLabel == 0) {
+        --GraphNodes[j].NumInEdges;
+        continue;
+      }
+
+      *(N->PointsTo) |= GraphNodes[j].PointsTo;
+
+      // If we didn't end up storing this in the hash, and we're done with all
+      // the edges, we don't need the points-to set anymore.
+      --GraphNodes[j].NumInEdges;
+      if (!GraphNodes[j].NumInEdges && !GraphNodes[j].StoredInHash) {
+        delete GraphNodes[j].PointsTo;
+        GraphNodes[j].PointsTo = NULL;
+      }
+    }
+  // If this isn't a direct node, generate a fresh variable.
+  if (!N->Direct) {
+    N->PointsTo->set(FirstRefNode + NodeIndex);
+  }
+
+  // See If we have something equivalent to us, if not, generate a new
+  // equivalence class.
+  if (N->PointsTo->empty()) {
+    delete N->PointsTo;
+    N->PointsTo = NULL;
+  } else {
+    if (N->Direct) {
+      N->PointerEquivLabel = Set2PEClass[N->PointsTo];
+      if (N->PointerEquivLabel == 0) {
+        unsigned EquivClass = PEClass++;
+        N->StoredInHash = true;
+        Set2PEClass[N->PointsTo] = EquivClass;
+        N->PointerEquivLabel = EquivClass;
+      }
+    } else {
+      N->PointerEquivLabel = PEClass++;
+    }
+  }
+}
+
+/// Rewrite our list of constraints so that pointer equivalent nodes are
+/// replaced by their the pointer equivalence class representative.
+void Andersens::RewriteConstraints() {
+  std::vector<Constraint> NewConstraints;
+  DenseSet<Constraint, ConstraintKeyInfo> Seen;
+
+  PEClass2Node.clear();
+  PENLEClass2Node.clear();
+
+  // We may have from 1 to Graphnodes + 1 equivalence classes.
+  PEClass2Node.insert(PEClass2Node.begin(), GraphNodes.size() + 1, -1);
+  PENLEClass2Node.insert(PENLEClass2Node.begin(), GraphNodes.size() + 1, -1);
+
+  // Rewrite constraints, ignoring non-pointer constraints, uniting equivalent
+  // nodes, and rewriting constraints to use the representative nodes.
+  for (unsigned i = 0, e = Constraints.size(); i != e; ++i) {
+    Constraint &C = Constraints[i];
+    unsigned RHSNode = FindNode(C.Src);
+    unsigned LHSNode = FindNode(C.Dest);
+    unsigned RHSLabel = GraphNodes[VSSCCRep[RHSNode]].PointerEquivLabel;
+    unsigned LHSLabel = GraphNodes[VSSCCRep[LHSNode]].PointerEquivLabel;
+
+    // First we try to eliminate constraints for things we can prove don't point
+    // to anything.
+    if (LHSLabel == 0) {
+      DEBUG(PrintNode(&GraphNodes[LHSNode]));
+      DOUT << " is a non-pointer, ignoring constraint.\n";
+      continue;
+    }
+    if (RHSLabel == 0) {
+      DEBUG(PrintNode(&GraphNodes[RHSNode]));
+      DOUT << " is a non-pointer, ignoring constraint.\n";
+      continue;
+    }
+    // This constraint may be useless, and it may become useless as we translate
+    // it.
+    if (C.Src == C.Dest && C.Type == Constraint::Copy)
+      continue;
+
+    C.Src = FindEquivalentNode(RHSNode, RHSLabel);
+    C.Dest = FindEquivalentNode(FindNode(LHSNode), LHSLabel);
+    if ((C.Src == C.Dest && C.Type == Constraint::Copy)
+        || Seen.count(C))
+      continue;
+
+    Seen.insert(C);
+    NewConstraints.push_back(C);
+  }
+  Constraints.swap(NewConstraints);
+  PEClass2Node.clear();
+}
+
+/// See if we have a node that is pointer equivalent to the one being asked
+/// about, and if so, unite them and return the equivalent node.  Otherwise,
+/// return the original node.
+unsigned Andersens::FindEquivalentNode(unsigned NodeIndex,
+                                       unsigned NodeLabel) {
+  if (!GraphNodes[NodeIndex].AddressTaken) {
+    if (PEClass2Node[NodeLabel] != -1) {
+      // We found an existing node with the same pointer label, so unify them.
+      // We specifically request that Union-By-Rank not be used so that
+      // PEClass2Node[NodeLabel] U= NodeIndex and not the other way around.
+      return UniteNodes(PEClass2Node[NodeLabel], NodeIndex, false);
+    } else {
+      PEClass2Node[NodeLabel] = NodeIndex;
+      PENLEClass2Node[NodeLabel] = NodeIndex;
+    }
+  } else if (PENLEClass2Node[NodeLabel] == -1) {
+    PENLEClass2Node[NodeLabel] = NodeIndex;
+  }
+
+  return NodeIndex;
+}
+
+void Andersens::PrintLabels() const {
+  for (unsigned i = 0; i < GraphNodes.size(); ++i) {
+    if (i < FirstRefNode) {
+      PrintNode(&GraphNodes[i]);
+    } else if (i < FirstAdrNode) {
+      DOUT << "REF(";
+      PrintNode(&GraphNodes[i-FirstRefNode]);
+      DOUT <<")";
+    } else {
+      DOUT << "ADR(";
+      PrintNode(&GraphNodes[i-FirstAdrNode]);
+      DOUT <<")";
+    }
+
+    DOUT << " has pointer label " << GraphNodes[i].PointerEquivLabel
+         << " and SCC rep " << VSSCCRep[i]
+         << " and is " << (GraphNodes[i].Direct ? "Direct" : "Not direct")
+         << "\n";
+  }
+}
+
+/// The technique used here is described in "The Ant and the
+/// Grasshopper: Fast and Accurate Pointer Analysis for Millions of
+/// Lines of Code. In Programming Language Design and Implementation
+/// (PLDI), June 2007." It is known as the "HCD" (Hybrid Cycle
+/// Detection) algorithm. It is called a hybrid because it performs an
+/// offline analysis and uses its results during the solving (online)
+/// phase. This is just the offline portion; the results of this
+/// operation are stored in SDT and are later used in SolveContraints()
+/// and UniteNodes().
+void Andersens::HCD() {
+  DOUT << "Starting HCD.\n";
+  HCDSCCRep.resize(GraphNodes.size());
+
+  for (unsigned i = 0; i < GraphNodes.size(); ++i) {
+    GraphNodes[i].Edges = new SparseBitVector<>;
+    HCDSCCRep[i] = i;
+  }
+
+  for (unsigned i = 0, e = Constraints.size(); i != e; ++i) {
+    Constraint &C = Constraints[i];
+    assert (C.Src < GraphNodes.size() && C.Dest < GraphNodes.size());
+    if (C.Type == Constraint::AddressOf) {
+      continue;
+    } else if (C.Type == Constraint::Load) {
+      if( C.Offset == 0 )
+        GraphNodes[C.Dest].Edges->set(C.Src + FirstRefNode);
+    } else if (C.Type == Constraint::Store) {
+      if( C.Offset == 0 )
+        GraphNodes[C.Dest + FirstRefNode].Edges->set(C.Src);
+    } else {
+      GraphNodes[C.Dest].Edges->set(C.Src);
+    }
+  }
+
+  Node2DFS.insert(Node2DFS.begin(), GraphNodes.size(), 0);
+  Node2Deleted.insert(Node2Deleted.begin(), GraphNodes.size(), false);
+  Node2Visited.insert(Node2Visited.begin(), GraphNodes.size(), false);
+  SDT.insert(SDT.begin(), GraphNodes.size() / 2, -1);
+
+  DFSNumber = 0;
+  for (unsigned i = 0; i < GraphNodes.size(); ++i) {
+    unsigned Node = HCDSCCRep[i];
+    if (!Node2Deleted[Node])
+      Search(Node);
+  }
+
+  for (unsigned i = 0; i < GraphNodes.size(); ++i)
+    if (GraphNodes[i].Edges != NULL) {
+      delete GraphNodes[i].Edges;
+      GraphNodes[i].Edges = NULL;
+    }
+
+  while( !SCCStack.empty() )
+    SCCStack.pop();
+
+  Node2DFS.clear();
+  Node2Visited.clear();
+  Node2Deleted.clear();
+  HCDSCCRep.clear();
+  DOUT << "HCD complete.\n";
+}
+
+// Component of HCD: 
+// Use Nuutila's variant of Tarjan's algorithm to detect
+// Strongly-Connected Components (SCCs). For non-trivial SCCs
+// containing ref nodes, insert the appropriate information in SDT.
+void Andersens::Search(unsigned Node) {
+  unsigned MyDFS = DFSNumber++;
+
+  Node2Visited[Node] = true;
+  Node2DFS[Node] = MyDFS;
+
+  for (SparseBitVector<>::iterator Iter = GraphNodes[Node].Edges->begin(),
+                                   End  = GraphNodes[Node].Edges->end();
+       Iter != End;
+       ++Iter) {
+    unsigned J = HCDSCCRep[*Iter];
+    assert(GraphNodes[J].isRep() && "Debug check; must be representative");
+    if (!Node2Deleted[J]) {
+      if (!Node2Visited[J])
+        Search(J);
+      if (Node2DFS[Node] > Node2DFS[J])
+        Node2DFS[Node] = Node2DFS[J];
+    }
+  }
+
+  if( MyDFS != Node2DFS[Node] ) {
+    SCCStack.push(Node);
+    return;
+  }
+
+  // This node is the root of a SCC, so process it.
+  //
+  // If the SCC is "non-trivial" (not a singleton) and contains a reference 
+  // node, we place this SCC into SDT.  We unite the nodes in any case.
+  if (!SCCStack.empty() && Node2DFS[SCCStack.top()] >= MyDFS) {
+    SparseBitVector<> SCC;
+
+    SCC.set(Node);
+
+    bool Ref = (Node >= FirstRefNode);
+
+    Node2Deleted[Node] = true;
+
+    do {
+      unsigned P = SCCStack.top(); SCCStack.pop();
+      Ref |= (P >= FirstRefNode);
+      SCC.set(P);
+      HCDSCCRep[P] = Node;
+    } while (!SCCStack.empty() && Node2DFS[SCCStack.top()] >= MyDFS);
+
+    if (Ref) {
+      unsigned Rep = SCC.find_first();
+      assert(Rep < FirstRefNode && "The SCC didn't have a non-Ref node!");
+
+      SparseBitVector<>::iterator i = SCC.begin();
+
+      // Skip over the non-ref nodes
+      while( *i < FirstRefNode )
+        ++i;
+
+      while( i != SCC.end() )
+        SDT[ (*i++) - FirstRefNode ] = Rep;
+    }
+  }
+}
+
+
+/// Optimize the constraints by performing offline variable substitution and
+/// other optimizations.
+void Andersens::OptimizeConstraints() {
+  DOUT << "Beginning constraint optimization\n";
+
+  SDTActive = false;
+
+  // Function related nodes need to stay in the same relative position and can't
+  // be location equivalent.
+  for (std::map<unsigned, unsigned>::iterator Iter = MaxK.begin();
+       Iter != MaxK.end();
+       ++Iter) {
+    for (unsigned i = Iter->first;
+         i != Iter->first + Iter->second;
+         ++i) {
+      GraphNodes[i].AddressTaken = true;
+      GraphNodes[i].Direct = false;
+    }
+  }
+
+  ClumpAddressTaken();
+  FirstRefNode = GraphNodes.size();
+  FirstAdrNode = FirstRefNode + GraphNodes.size();
+  GraphNodes.insert(GraphNodes.end(), 2 * GraphNodes.size(),
+                    Node(false));
+  VSSCCRep.resize(GraphNodes.size());
+  for (unsigned i = 0; i < GraphNodes.size(); ++i) {
+    VSSCCRep[i] = i;
+  }
+  HVN();
+  for (unsigned i = 0; i < GraphNodes.size(); ++i) {
+    Node *N = &GraphNodes[i];
+    delete N->PredEdges;
+    N->PredEdges = NULL;
+    delete N->ImplicitPredEdges;
+    N->ImplicitPredEdges = NULL;
+  }
+#undef DEBUG_TYPE
+#define DEBUG_TYPE "anders-aa-labels"
+  DEBUG(PrintLabels());
+#undef DEBUG_TYPE
+#define DEBUG_TYPE "anders-aa"
+  RewriteConstraints();
+  // Delete the adr nodes.
+  GraphNodes.resize(FirstRefNode * 2);
+
+  // Now perform HU
+  for (unsigned i = 0; i < GraphNodes.size(); ++i) {
+    Node *N = &GraphNodes[i];
+    if (FindNode(i) == i) {
+      N->PointsTo = new SparseBitVector<>;
+      N->PointedToBy = new SparseBitVector<>;
+      // Reset our labels
+    }
+    VSSCCRep[i] = i;
+    N->PointerEquivLabel = 0;
+  }
+  HU();
+#undef DEBUG_TYPE
+#define DEBUG_TYPE "anders-aa-labels"
+  DEBUG(PrintLabels());
+#undef DEBUG_TYPE
+#define DEBUG_TYPE "anders-aa"
+  RewriteConstraints();
+  for (unsigned i = 0; i < GraphNodes.size(); ++i) {
+    if (FindNode(i) == i) {
+      Node *N = &GraphNodes[i];
+      delete N->PointsTo;
+      N->PointsTo = NULL;
+      delete N->PredEdges;
+      N->PredEdges = NULL;
+      delete N->ImplicitPredEdges;
+      N->ImplicitPredEdges = NULL;
+      delete N->PointedToBy;
+      N->PointedToBy = NULL;
+    }
+  }
+
+  // perform Hybrid Cycle Detection (HCD)
+  HCD();
+  SDTActive = true;
+
+  // No longer any need for the upper half of GraphNodes (for ref nodes).
+  GraphNodes.erase(GraphNodes.begin() + FirstRefNode, GraphNodes.end());
+
+  // HCD complete.
+
+  DOUT << "Finished constraint optimization\n";
+  FirstRefNode = 0;
+  FirstAdrNode = 0;
+}
+
+/// Unite pointer but not location equivalent variables, now that the constraint
+/// graph is built.
+void Andersens::UnitePointerEquivalences() {
+  DOUT << "Uniting remaining pointer equivalences\n";
+  for (unsigned i = 0; i < GraphNodes.size(); ++i) {
+    if (GraphNodes[i].AddressTaken && GraphNodes[i].isRep()) {
+      unsigned Label = GraphNodes[i].PointerEquivLabel;
+
+      if (Label && PENLEClass2Node[Label] != -1)
+        UniteNodes(i, PENLEClass2Node[Label]);
+    }
+  }
+  DOUT << "Finished remaining pointer equivalences\n";
+  PENLEClass2Node.clear();
+}
+
+/// Create the constraint graph used for solving points-to analysis.
+///
+void Andersens::CreateConstraintGraph() {
+  for (unsigned i = 0, e = Constraints.size(); i != e; ++i) {
+    Constraint &C = Constraints[i];
+    assert (C.Src < GraphNodes.size() && C.Dest < GraphNodes.size());
+    if (C.Type == Constraint::AddressOf)
+      GraphNodes[C.Dest].PointsTo->set(C.Src);
+    else if (C.Type == Constraint::Load)
+      GraphNodes[C.Src].Constraints.push_back(C);
+    else if (C.Type == Constraint::Store)
+      GraphNodes[C.Dest].Constraints.push_back(C);
+    else if (C.Offset != 0)
+      GraphNodes[C.Src].Constraints.push_back(C);
+    else
+      GraphNodes[C.Src].Edges->set(C.Dest);
+  }
+}
+
+// Perform DFS and cycle detection.
+bool Andersens::QueryNode(unsigned Node) {
+  assert(GraphNodes[Node].isRep() && "Querying a non-rep node");
+  unsigned OurDFS = ++DFSNumber;
+  SparseBitVector<> ToErase;
+  SparseBitVector<> NewEdges;
+  Tarjan2DFS[Node] = OurDFS;
+
+  // Changed denotes a change from a recursive call that we will bubble up.
+  // Merged is set if we actually merge a node ourselves.
+  bool Changed = false, Merged = false;
+
+  for (SparseBitVector<>::iterator bi = GraphNodes[Node].Edges->begin();
+       bi != GraphNodes[Node].Edges->end();
+       ++bi) {
+    unsigned RepNode = FindNode(*bi);
+    // If this edge points to a non-representative node but we are
+    // already planning to add an edge to its representative, we have no
+    // need for this edge anymore.
+    if (RepNode != *bi && NewEdges.test(RepNode)){
+      ToErase.set(*bi);
+      continue;
+    }
+
+    // Continue about our DFS.
+    if (!Tarjan2Deleted[RepNode]){
+      if (Tarjan2DFS[RepNode] == 0) {
+        Changed |= QueryNode(RepNode);
+        // May have been changed by QueryNode
+        RepNode = FindNode(RepNode);
+      }
+      if (Tarjan2DFS[RepNode] < Tarjan2DFS[Node])
+        Tarjan2DFS[Node] = Tarjan2DFS[RepNode];
+    }
+
+    // We may have just discovered that this node is part of a cycle, in
+    // which case we can also erase it.
+    if (RepNode != *bi) {
+      ToErase.set(*bi);
+      NewEdges.set(RepNode);
+    }
+  }
+
+  GraphNodes[Node].Edges->intersectWithComplement(ToErase);
+  GraphNodes[Node].Edges |= NewEdges;
+
+  // If this node is a root of a non-trivial SCC, place it on our 
+  // worklist to be processed.
+  if (OurDFS == Tarjan2DFS[Node]) {
+    while (!SCCStack.empty() && Tarjan2DFS[SCCStack.top()] >= OurDFS) {
+      Node = UniteNodes(Node, SCCStack.top());
+
+      SCCStack.pop();
+      Merged = true;
+    }
+    Tarjan2Deleted[Node] = true;
+
+    if (Merged)
+      NextWL->insert(&GraphNodes[Node]);
+  } else {
+    SCCStack.push(Node);
+  }
+
+  return(Changed | Merged);
+}
+
+/// SolveConstraints - This stage iteratively processes the constraints list
+/// propagating constraints (adding edges to the Nodes in the points-to graph)
+/// until a fixed point is reached.
+///
+/// We use a variant of the technique called "Lazy Cycle Detection", which is
+/// described in "The Ant and the Grasshopper: Fast and Accurate Pointer
+/// Analysis for Millions of Lines of Code. In Programming Language Design and
+/// Implementation (PLDI), June 2007."
+/// The paper describes performing cycle detection one node at a time, which can
+/// be expensive if there are no cycles, but there are long chains of nodes that
+/// it heuristically believes are cycles (because it will DFS from each node
+/// without state from previous nodes).
+/// Instead, we use the heuristic to build a worklist of nodes to check, then
+/// cycle detect them all at the same time to do this more cheaply.  This
+/// catches cycles slightly later than the original technique did, but does it
+/// make significantly cheaper.
+
+void Andersens::SolveConstraints() {
+  CurrWL = &w1;
+  NextWL = &w2;
+
+  OptimizeConstraints();
+#undef DEBUG_TYPE
+#define DEBUG_TYPE "anders-aa-constraints"
+      DEBUG(PrintConstraints());
+#undef DEBUG_TYPE
+#define DEBUG_TYPE "anders-aa"
+
+  for (unsigned i = 0; i < GraphNodes.size(); ++i) {
+    Node *N = &GraphNodes[i];
+    N->PointsTo = new SparseBitVector<>;
+    N->OldPointsTo = new SparseBitVector<>;
+    N->Edges = new SparseBitVector<>;
+  }
+  CreateConstraintGraph();
+  UnitePointerEquivalences();
+  assert(SCCStack.empty() && "SCC Stack should be empty by now!");
+  Node2DFS.clear();
+  Node2Deleted.clear();
+  Node2DFS.insert(Node2DFS.begin(), GraphNodes.size(), 0);
+  Node2Deleted.insert(Node2Deleted.begin(), GraphNodes.size(), false);
+  DFSNumber = 0;
+  DenseSet<Constraint, ConstraintKeyInfo> Seen;
+  DenseSet<std::pair<unsigned,unsigned>, PairKeyInfo> EdgesChecked;
+
+  // Order graph and add initial nodes to work list.
+  for (unsigned i = 0; i < GraphNodes.size(); ++i) {
+    Node *INode = &GraphNodes[i];
+
+    // Add to work list if it's a representative and can contribute to the
+    // calculation right now.
+    if (INode->isRep() && !INode->PointsTo->empty()
+        && (!INode->Edges->empty() || !INode->Constraints.empty())) {
+      INode->Stamp();
+      CurrWL->insert(INode);
+    }
+  }
+  std::queue<unsigned int> TarjanWL;
+#if !FULL_UNIVERSAL
+  // "Rep and special variables" - in order for HCD to maintain conservative
+  // results when !FULL_UNIVERSAL, we need to treat the special variables in
+  // the same way that the !FULL_UNIVERSAL tweak does throughout the rest of
+  // the analysis - it's ok to add edges from the special nodes, but never
+  // *to* the special nodes.
+  std::vector<unsigned int> RSV;
+#endif
+  while( !CurrWL->empty() ) {
+    DOUT << "Starting iteration #" << ++NumIters << "\n";
+
+    Node* CurrNode;
+    unsigned CurrNodeIndex;
+
+    // Actual cycle checking code.  We cycle check all of the lazy cycle
+    // candidates from the last iteration in one go.
+    if (!TarjanWL.empty()) {
+      DFSNumber = 0;
+      
+      Tarjan2DFS.clear();
+      Tarjan2Deleted.clear();
+      while (!TarjanWL.empty()) {
+        unsigned int ToTarjan = TarjanWL.front();
+        TarjanWL.pop();
+        if (!Tarjan2Deleted[ToTarjan]
+            && GraphNodes[ToTarjan].isRep()
+            && Tarjan2DFS[ToTarjan] == 0)
+          QueryNode(ToTarjan);
+      }
+    }
+    
+    // Add to work list if it's a representative and can contribute to the
+    // calculation right now.
+    while( (CurrNode = CurrWL->pop()) != NULL ) {
+      CurrNodeIndex = CurrNode - &GraphNodes[0];
+      CurrNode->Stamp();
+      
+          
+      // Figure out the changed points to bits
+      SparseBitVector<> CurrPointsTo;
+      CurrPointsTo.intersectWithComplement(CurrNode->PointsTo,
+                                           CurrNode->OldPointsTo);
+      if (CurrPointsTo.empty())
+        continue;
+
+      *(CurrNode->OldPointsTo) |= CurrPointsTo;
+
+      // Check the offline-computed equivalencies from HCD.
+      bool SCC = false;
+      unsigned Rep;
+
+      if (SDT[CurrNodeIndex] >= 0) {
+        SCC = true;
+        Rep = FindNode(SDT[CurrNodeIndex]);
+
+#if !FULL_UNIVERSAL
+        RSV.clear();
+#endif
+        for (SparseBitVector<>::iterator bi = CurrPointsTo.begin();
+             bi != CurrPointsTo.end(); ++bi) {
+          unsigned Node = FindNode(*bi);
+#if !FULL_UNIVERSAL
+          if (Node < NumberSpecialNodes) {
+            RSV.push_back(Node);
+            continue;
+          }
+#endif
+          Rep = UniteNodes(Rep,Node);
+        }
+#if !FULL_UNIVERSAL
+        RSV.push_back(Rep);
+#endif
+
+        NextWL->insert(&GraphNodes[Rep]);
+
+        if ( ! CurrNode->isRep() )
+          continue;
+      }
+
+      Seen.clear();
+
+      /* Now process the constraints for this node.  */
+      for (std::list<Constraint>::iterator li = CurrNode->Constraints.begin();
+           li != CurrNode->Constraints.end(); ) {
+        li->Src = FindNode(li->Src);
+        li->Dest = FindNode(li->Dest);
+
+        // Delete redundant constraints
+        if( Seen.count(*li) ) {
+          std::list<Constraint>::iterator lk = li; li++;
+
+          CurrNode->Constraints.erase(lk);
+          ++NumErased;
+          continue;
+        }
+        Seen.insert(*li);
+
+        // Src and Dest will be the vars we are going to process.
+        // This may look a bit ugly, but what it does is allow us to process
+        // both store and load constraints with the same code.
+        // Load constraints say that every member of our RHS solution has K
+        // added to it, and that variable gets an edge to LHS. We also union
+        // RHS+K's solution into the LHS solution.
+        // Store constraints say that every member of our LHS solution has K
+        // added to it, and that variable gets an edge from RHS. We also union
+        // RHS's solution into the LHS+K solution.
+        unsigned *Src;
+        unsigned *Dest;
+        unsigned K = li->Offset;
+        unsigned CurrMember;
+        if (li->Type == Constraint::Load) {
+          Src = &CurrMember;
+          Dest = &li->Dest;
+        } else if (li->Type == Constraint::Store) {
+          Src = &li->Src;
+          Dest = &CurrMember;
+        } else {
+          // TODO Handle offseted copy constraint
+          li++;
+          continue;
+        }
+
+        // See if we can use Hybrid Cycle Detection (that is, check
+        // if it was a statically detected offline equivalence that
+        // involves pointers; if so, remove the redundant constraints).
+        if( SCC && K == 0 ) {
+#if FULL_UNIVERSAL
+          CurrMember = Rep;
+
+          if (GraphNodes[*Src].Edges->test_and_set(*Dest))
+            if (GraphNodes[*Dest].PointsTo |= *(GraphNodes[*Src].PointsTo))
+              NextWL->insert(&GraphNodes[*Dest]);
+#else
+          for (unsigned i=0; i < RSV.size(); ++i) {
+            CurrMember = RSV[i];
+
+            if (*Dest < NumberSpecialNodes)
+              continue;
+            if (GraphNodes[*Src].Edges->test_and_set(*Dest))
+              if (GraphNodes[*Dest].PointsTo |= *(GraphNodes[*Src].PointsTo))
+                NextWL->insert(&GraphNodes[*Dest]);
+          }
+#endif
+          // since all future elements of the points-to set will be
+          // equivalent to the current ones, the complex constraints
+          // become redundant.
+          //
+          std::list<Constraint>::iterator lk = li; li++;
+#if !FULL_UNIVERSAL
+          // In this case, we can still erase the constraints when the
+          // elements of the points-to sets are referenced by *Dest,
+          // but not when they are referenced by *Src (i.e. for a Load
+          // constraint). This is because if another special variable is
+          // put into the points-to set later, we still need to add the
+          // new edge from that special variable.
+          if( lk->Type != Constraint::Load)
+#endif
+          GraphNodes[CurrNodeIndex].Constraints.erase(lk);
+        } else {
+          const SparseBitVector<> &Solution = CurrPointsTo;
+
+          for (SparseBitVector<>::iterator bi = Solution.begin();
+               bi != Solution.end();
+               ++bi) {
+            CurrMember = *bi;
+
+            // Need to increment the member by K since that is where we are
+            // supposed to copy to/from.  Note that in positive weight cycles,
+            // which occur in address taking of fields, K can go past
+            // MaxK[CurrMember] elements, even though that is all it could point
+            // to.
+            if (K > 0 && K > MaxK[CurrMember])
+              continue;
+            else
+              CurrMember = FindNode(CurrMember + K);
+
+            // Add an edge to the graph, so we can just do regular
+            // bitmap ior next time.  It may also let us notice a cycle.
+#if !FULL_UNIVERSAL
+            if (*Dest < NumberSpecialNodes)
+              continue;
+#endif
+            if (GraphNodes[*Src].Edges->test_and_set(*Dest))
+              if (GraphNodes[*Dest].PointsTo |= *(GraphNodes[*Src].PointsTo))
+                NextWL->insert(&GraphNodes[*Dest]);
+
+          }
+          li++;
+        }
+      }
+      SparseBitVector<> NewEdges;
+      SparseBitVector<> ToErase;
+
+      // Now all we have left to do is propagate points-to info along the
+      // edges, erasing the redundant edges.
+      for (SparseBitVector<>::iterator bi = CurrNode->Edges->begin();
+           bi != CurrNode->Edges->end();
+           ++bi) {
+
+        unsigned DestVar = *bi;
+        unsigned Rep = FindNode(DestVar);
+
+        // If we ended up with this node as our destination, or we've already
+        // got an edge for the representative, delete the current edge.
+        if (Rep == CurrNodeIndex ||
+            (Rep != DestVar && NewEdges.test(Rep))) {
+            ToErase.set(DestVar);
+            continue;
+        }
+        
+        std::pair<unsigned,unsigned> edge(CurrNodeIndex,Rep);
+        
+        // This is where we do lazy cycle detection.
+        // If this is a cycle candidate (equal points-to sets and this
+        // particular edge has not been cycle-checked previously), add to the
+        // list to check for cycles on the next iteration.
+        if (!EdgesChecked.count(edge) &&
+            *(GraphNodes[Rep].PointsTo) == *(CurrNode->PointsTo)) {
+          EdgesChecked.insert(edge);
+          TarjanWL.push(Rep);
+        }
+        // Union the points-to sets into the dest
+#if !FULL_UNIVERSAL
+        if (Rep >= NumberSpecialNodes)
+#endif
+        if (GraphNodes[Rep].PointsTo |= CurrPointsTo) {
+          NextWL->insert(&GraphNodes[Rep]);
+        }
+        // If this edge's destination was collapsed, rewrite the edge.
+        if (Rep != DestVar) {
+          ToErase.set(DestVar);
+          NewEdges.set(Rep);
+        }
+      }
+      CurrNode->Edges->intersectWithComplement(ToErase);
+      CurrNode->Edges |= NewEdges;
+    }
+
+    // Switch to other work list.
+    WorkList* t = CurrWL; CurrWL = NextWL; NextWL = t;
+  }
+
+
+  Node2DFS.clear();
+  Node2Deleted.clear();
+  for (unsigned i = 0; i < GraphNodes.size(); ++i) {
+    Node *N = &GraphNodes[i];
+    delete N->OldPointsTo;
+    delete N->Edges;
+  }
+  SDTActive = false;
+  SDT.clear();
+}
+
+//===----------------------------------------------------------------------===//
+//                               Union-Find
+//===----------------------------------------------------------------------===//
+
+// Unite nodes First and Second, returning the one which is now the
+// representative node.  First and Second are indexes into GraphNodes
+unsigned Andersens::UniteNodes(unsigned First, unsigned Second,
+                               bool UnionByRank) {
+  assert (First < GraphNodes.size() && Second < GraphNodes.size() &&
+          "Attempting to merge nodes that don't exist");
+
+  Node *FirstNode = &GraphNodes[First];
+  Node *SecondNode = &GraphNodes[Second];
+
+  assert (SecondNode->isRep() && FirstNode->isRep() &&
+          "Trying to unite two non-representative nodes!");
+  if (First == Second)
+    return First;
+
+  if (UnionByRank) {
+    int RankFirst  = (int) FirstNode ->NodeRep;
+    int RankSecond = (int) SecondNode->NodeRep;
+
+    // Rank starts at -1 and gets decremented as it increases.
+    // Translation: higher rank, lower NodeRep value, which is always negative.
+    if (RankFirst > RankSecond) {
+      unsigned t = First; First = Second; Second = t;
+      Node* tp = FirstNode; FirstNode = SecondNode; SecondNode = tp;
+    } else if (RankFirst == RankSecond) {
+      FirstNode->NodeRep = (unsigned) (RankFirst - 1);
+    }
+  }
+
+  SecondNode->NodeRep = First;
+#if !FULL_UNIVERSAL
+  if (First >= NumberSpecialNodes)
+#endif
+  if (FirstNode->PointsTo && SecondNode->PointsTo)
+    FirstNode->PointsTo |= *(SecondNode->PointsTo);
+  if (FirstNode->Edges && SecondNode->Edges)
+    FirstNode->Edges |= *(SecondNode->Edges);
+  if (!SecondNode->Constraints.empty())
+    FirstNode->Constraints.splice(FirstNode->Constraints.begin(),
+                                  SecondNode->Constraints);
+  if (FirstNode->OldPointsTo) {
+    delete FirstNode->OldPointsTo;
+    FirstNode->OldPointsTo = new SparseBitVector<>;
+  }
+
+  // Destroy interesting parts of the merged-from node.
+  delete SecondNode->OldPointsTo;
+  delete SecondNode->Edges;
+  delete SecondNode->PointsTo;
+  SecondNode->Edges = NULL;
+  SecondNode->PointsTo = NULL;
+  SecondNode->OldPointsTo = NULL;
+
+  NumUnified++;
+  DOUT << "Unified Node ";
+  DEBUG(PrintNode(FirstNode));
+  DOUT << " and Node ";
+  DEBUG(PrintNode(SecondNode));
+  DOUT << "\n";
+
+  if (SDTActive)
+    if (SDT[Second] >= 0) {
+      if (SDT[First] < 0)
+        SDT[First] = SDT[Second];
+      else {
+        UniteNodes( FindNode(SDT[First]), FindNode(SDT[Second]) );
+        First = FindNode(First);
+      }
+    }
+
+  return First;
+}
+
+// Find the index into GraphNodes of the node representing Node, performing
+// path compression along the way
+unsigned Andersens::FindNode(unsigned NodeIndex) {
+  assert (NodeIndex < GraphNodes.size()
+          && "Attempting to find a node that can't exist");
+  Node *N = &GraphNodes[NodeIndex];
+  if (N->isRep())
+    return NodeIndex;
+  else
+    return (N->NodeRep = FindNode(N->NodeRep));
+}
+
+// Find the index into GraphNodes of the node representing Node, 
+// don't perform path compression along the way (for Print)
+unsigned Andersens::FindNode(unsigned NodeIndex) const {
+  assert (NodeIndex < GraphNodes.size()
+          && "Attempting to find a node that can't exist");
+  const Node *N = &GraphNodes[NodeIndex];
+  if (N->isRep())
+    return NodeIndex;
+  else
+    return FindNode(N->NodeRep);
+}
+
+//===----------------------------------------------------------------------===//
+//                               Debugging Output
+//===----------------------------------------------------------------------===//
+
+void Andersens::PrintNode(const Node *N) const {
+  if (N == &GraphNodes[UniversalSet]) {
+    cerr << "<universal>";
+    return;
+  } else if (N == &GraphNodes[NullPtr]) {
+    cerr << "<nullptr>";
+    return;
+  } else if (N == &GraphNodes[NullObject]) {
+    cerr << "<null>";
+    return;
+  }
+  if (!N->getValue()) {
+    cerr << "artificial" << (intptr_t) N;
+    return;
+  }
+
+  assert(N->getValue() != 0 && "Never set node label!");
+  Value *V = N->getValue();
+  if (Function *F = dyn_cast<Function>(V)) {
+    if (isa<PointerType>(F->getFunctionType()->getReturnType()) &&
+        N == &GraphNodes[getReturnNode(F)]) {
+      cerr << F->getName() << ":retval";
+      return;
+    } else if (F->getFunctionType()->isVarArg() &&
+               N == &GraphNodes[getVarargNode(F)]) {
+      cerr << F->getName() << ":vararg";
+      return;
+    }
+  }
+
+  if (Instruction *I = dyn_cast<Instruction>(V))
+    cerr << I->getParent()->getParent()->getName() << ":";
+  else if (Argument *Arg = dyn_cast<Argument>(V))
+    cerr << Arg->getParent()->getName() << ":";
+
+  if (V->hasName())
+    cerr << V->getName();
+  else
+    cerr << "(unnamed)";
+
+  if (isa<GlobalValue>(V) || isa<AllocationInst>(V))
+    if (N == &GraphNodes[getObject(V)])
+      cerr << "<mem>";
+}
+void Andersens::PrintConstraint(const Constraint &C) const {
+  if (C.Type == Constraint::Store) {
+    cerr << "*";
+    if (C.Offset != 0)
+      cerr << "(";
+  }
+  PrintNode(&GraphNodes[C.Dest]);
+  if (C.Type == Constraint::Store && C.Offset != 0)
+    cerr << " + " << C.Offset << ")";
+  cerr << " = ";
+  if (C.Type == Constraint::Load) {
+    cerr << "*";
+    if (C.Offset != 0)
+      cerr << "(";
+  }
+  else if (C.Type == Constraint::AddressOf)
+    cerr << "&";
+  PrintNode(&GraphNodes[C.Src]);
+  if (C.Offset != 0 && C.Type != Constraint::Store)
+    cerr << " + " << C.Offset;
+  if (C.Type == Constraint::Load && C.Offset != 0)
+    cerr << ")";
+  cerr << "\n";
+}
+
+void Andersens::PrintConstraints() const {
+  cerr << "Constraints:\n";
+
+  for (unsigned i = 0, e = Constraints.size(); i != e; ++i)
+    PrintConstraint(Constraints[i]);
+}
+
+void Andersens::PrintPointsToGraph() const {
+  cerr << "Points-to graph:\n";
+  for (unsigned i = 0, e = GraphNodes.size(); i != e; ++i) {
+    const Node *N = &GraphNodes[i];
+    if (FindNode(i) != i) {
+      PrintNode(N);
+      cerr << "\t--> same as ";
+      PrintNode(&GraphNodes[FindNode(i)]);
+      cerr << "\n";
+    } else {
+      cerr << "[" << (N->PointsTo->count()) << "] ";
+      PrintNode(N);
+      cerr << "\t--> ";
+
+      bool first = true;
+      for (SparseBitVector<>::iterator bi = N->PointsTo->begin();
+           bi != N->PointsTo->end();
+           ++bi) {
+        if (!first)
+          cerr << ", ";
+        PrintNode(&GraphNodes[*bi]);
+        first = false;
+      }
+      cerr << "\n";
+    }
+  }
+}
diff --git a/lib/Analysis/IPA/CMakeLists.txt b/lib/Analysis/IPA/CMakeLists.txt
new file mode 100644
index 0000000..1ebb0be
--- /dev/null
+++ b/lib/Analysis/IPA/CMakeLists.txt
@@ -0,0 +1,7 @@
+add_llvm_library(LLVMipa
+  Andersens.cpp
+  CallGraph.cpp
+  CallGraphSCCPass.cpp
+  FindUsedTypes.cpp
+  GlobalsModRef.cpp
+  )
diff --git a/lib/Analysis/IPA/CallGraph.cpp b/lib/Analysis/IPA/CallGraph.cpp
new file mode 100644
index 0000000..6dabcdb
--- /dev/null
+++ b/lib/Analysis/IPA/CallGraph.cpp
@@ -0,0 +1,314 @@
+//===- CallGraph.cpp - Build a Module's call graph ------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This file implements the CallGraph class and provides the BasicCallGraph
+// default implementation.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Analysis/CallGraph.h"
+#include "llvm/Module.h"
+#include "llvm/Instructions.h"
+#include "llvm/IntrinsicInst.h"
+#include "llvm/Support/CallSite.h"
+#include "llvm/Support/Compiler.h"
+#include "llvm/Support/Streams.h"
+#include <ostream>
+using namespace llvm;
+
+namespace {
+
+//===----------------------------------------------------------------------===//
+// BasicCallGraph class definition
+//
+class VISIBILITY_HIDDEN BasicCallGraph : public CallGraph, public ModulePass {
+  // Root is root of the call graph, or the external node if a 'main' function
+  // couldn't be found.
+  //
+  CallGraphNode *Root;
+
+  // ExternalCallingNode - This node has edges to all external functions and
+  // those internal functions that have their address taken.
+  CallGraphNode *ExternalCallingNode;
+
+  // CallsExternalNode - This node has edges to it from all functions making
+  // indirect calls or calling an external function.
+  CallGraphNode *CallsExternalNode;
+
+public:
+  static char ID; // Class identification, replacement for typeinfo
+  BasicCallGraph() : ModulePass(&ID), Root(0), 
+    ExternalCallingNode(0), CallsExternalNode(0) {}
+
+  // runOnModule - Compute the call graph for the specified module.
+  virtual bool runOnModule(Module &M) {
+    CallGraph::initialize(M);
+    
+    ExternalCallingNode = getOrInsertFunction(0);
+    CallsExternalNode = new CallGraphNode(0);
+    Root = 0;
+  
+    // Add every function to the call graph...
+    for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I)
+      addToCallGraph(I);
+  
+    // If we didn't find a main function, use the external call graph node
+    if (Root == 0) Root = ExternalCallingNode;
+    
+    return false;
+  }
+
+  virtual void getAnalysisUsage(AnalysisUsage &AU) const {
+    AU.setPreservesAll();
+  }
+
+  void print(std::ostream *o, const Module *M) const {
+    if (o) print(*o, M);
+  }
+
+  virtual void print(std::ostream &o, const Module *M) const {
+    o << "CallGraph Root is: ";
+    if (Function *F = getRoot()->getFunction())
+      o << F->getName() << "\n";
+    else
+      o << "<<null function: 0x" << getRoot() << ">>\n";
+    
+    CallGraph::print(o, M);
+  }
+
+  virtual void releaseMemory() {
+    destroy();
+  }
+  
+  /// dump - Print out this call graph.
+  ///
+  inline void dump() const {
+    print(cerr, Mod);
+  }
+
+  CallGraphNode* getExternalCallingNode() const { return ExternalCallingNode; }
+  CallGraphNode* getCallsExternalNode()   const { return CallsExternalNode; }
+
+  // getRoot - Return the root of the call graph, which is either main, or if
+  // main cannot be found, the external node.
+  //
+  CallGraphNode *getRoot()             { return Root; }
+  const CallGraphNode *getRoot() const { return Root; }
+
+private:
+  //===---------------------------------------------------------------------
+  // Implementation of CallGraph construction
+  //
+
+  // addToCallGraph - Add a function to the call graph, and link the node to all
+  // of the functions that it calls.
+  //
+  void addToCallGraph(Function *F) {
+    CallGraphNode *Node = getOrInsertFunction(F);
+
+    // If this function has external linkage, anything could call it.
+    if (!F->hasLocalLinkage()) {
+      ExternalCallingNode->addCalledFunction(CallSite(), Node);
+
+      // Found the entry point?
+      if (F->getName() == "main") {
+        if (Root)    // Found multiple external mains?  Don't pick one.
+          Root = ExternalCallingNode;
+        else
+          Root = Node;          // Found a main, keep track of it!
+      }
+    }
+
+    // Loop over all of the users of the function, looking for non-call uses.
+    for (Value::use_iterator I = F->use_begin(), E = F->use_end(); I != E; ++I)
+      if ((!isa<CallInst>(I) && !isa<InvokeInst>(I))
+          || !CallSite(cast<Instruction>(I)).isCallee(I)) {
+        // Not a call, or being used as a parameter rather than as the callee.
+        ExternalCallingNode->addCalledFunction(CallSite(), Node);
+        break;
+      }
+
+    // If this function is not defined in this translation unit, it could call
+    // anything.
+    if (F->isDeclaration() && !F->isIntrinsic())
+      Node->addCalledFunction(CallSite(), CallsExternalNode);
+
+    // Look for calls by this function.
+    for (Function::iterator BB = F->begin(), BBE = F->end(); BB != BBE; ++BB)
+      for (BasicBlock::iterator II = BB->begin(), IE = BB->end();
+           II != IE; ++II) {
+        CallSite CS = CallSite::get(II);
+        if (CS.getInstruction() && !isa<DbgInfoIntrinsic>(II)) {
+          const Function *Callee = CS.getCalledFunction();
+          if (Callee)
+            Node->addCalledFunction(CS, getOrInsertFunction(Callee));
+          else
+            Node->addCalledFunction(CS, CallsExternalNode);
+        }
+      }
+  }
+
+  //
+  // destroy - Release memory for the call graph
+  virtual void destroy() {
+    /// CallsExternalNode is not in the function map, delete it explicitly.
+    delete CallsExternalNode;
+    CallsExternalNode = 0;
+    CallGraph::destroy();
+  }
+};
+
+} //End anonymous namespace
+
+static RegisterAnalysisGroup<CallGraph> X("Call Graph");
+static RegisterPass<BasicCallGraph>
+Y("basiccg", "Basic CallGraph Construction", false, true);
+static RegisterAnalysisGroup<CallGraph, true> Z(Y);
+
+char CallGraph::ID = 0;
+char BasicCallGraph::ID = 0;
+
+void CallGraph::initialize(Module &M) {
+  Mod = &M;
+}
+
+void CallGraph::destroy() {
+  if (!FunctionMap.empty()) {
+    for (FunctionMapTy::iterator I = FunctionMap.begin(), E = FunctionMap.end();
+        I != E; ++I)
+      delete I->second;
+    FunctionMap.clear();
+  }
+}
+
+void CallGraph::print(std::ostream &OS, const Module *M) const {
+  for (CallGraph::const_iterator I = begin(), E = end(); I != E; ++I)
+    I->second->print(OS);
+}
+
+void CallGraph::dump() const {
+  print(cerr, 0);
+}
+
+//===----------------------------------------------------------------------===//
+// Implementations of public modification methods
+//
+
+// removeFunctionFromModule - Unlink the function from this module, returning
+// it.  Because this removes the function from the module, the call graph node
+// is destroyed.  This is only valid if the function does not call any other
+// functions (ie, there are no edges in it's CGN).  The easiest way to do this
+// is to dropAllReferences before calling this.
+//
+Function *CallGraph::removeFunctionFromModule(CallGraphNode *CGN) {
+  assert(CGN->CalledFunctions.empty() && "Cannot remove function from call "
+         "graph if it references other functions!");
+  Function *F = CGN->getFunction(); // Get the function for the call graph node
+  delete CGN;                       // Delete the call graph node for this func
+  FunctionMap.erase(F);             // Remove the call graph node from the map
+
+  Mod->getFunctionList().remove(F);
+  return F;
+}
+
+// changeFunction - This method changes the function associated with this
+// CallGraphNode, for use by transformations that need to change the prototype
+// of a Function (thus they must create a new Function and move the old code
+// over).
+void CallGraph::changeFunction(Function *OldF, Function *NewF) {
+  iterator I = FunctionMap.find(OldF);
+  CallGraphNode *&New = FunctionMap[NewF];
+  assert(I != FunctionMap.end() && I->second && !New &&
+         "OldF didn't exist in CG or NewF already does!");
+  New = I->second;
+  New->F = NewF;
+  FunctionMap.erase(I);
+}
+
+// getOrInsertFunction - This method is identical to calling operator[], but
+// it will insert a new CallGraphNode for the specified function if one does
+// not already exist.
+CallGraphNode *CallGraph::getOrInsertFunction(const Function *F) {
+  CallGraphNode *&CGN = FunctionMap[F];
+  if (CGN) return CGN;
+  
+  assert((!F || F->getParent() == Mod) && "Function not in current module!");
+  return CGN = new CallGraphNode(const_cast<Function*>(F));
+}
+
+void CallGraphNode::print(std::ostream &OS) const {
+  if (Function *F = getFunction())
+    OS << "Call graph node for function: '" << F->getName() <<"'\n";
+  else
+    OS << "Call graph node <<null function: 0x" << this << ">>:\n";
+
+  for (const_iterator I = begin(), E = end(); I != E; ++I)
+    if (Function *FI = I->second->getFunction())
+      OS << "  Calls function '" << FI->getName() <<"'\n";
+  else
+    OS << "  Calls external node\n";
+  OS << "\n";
+}
+
+void CallGraphNode::dump() const { print(cerr); }
+
+/// removeCallEdgeFor - This method removes the edge in the node for the
+/// specified call site.  Note that this method takes linear time, so it
+/// should be used sparingly.
+void CallGraphNode::removeCallEdgeFor(CallSite CS) {
+  for (CalledFunctionsVector::iterator I = CalledFunctions.begin(); ; ++I) {
+    assert(I != CalledFunctions.end() && "Cannot find callsite to remove!");
+    if (I->first == CS) {
+      CalledFunctions.erase(I);
+      return;
+    }
+  }
+}
+
+
+// removeAnyCallEdgeTo - This method removes any call edges from this node to
+// the specified callee function.  This takes more time to execute than
+// removeCallEdgeTo, so it should not be used unless necessary.
+void CallGraphNode::removeAnyCallEdgeTo(CallGraphNode *Callee) {
+  for (unsigned i = 0, e = CalledFunctions.size(); i != e; ++i)
+    if (CalledFunctions[i].second == Callee) {
+      CalledFunctions[i] = CalledFunctions.back();
+      CalledFunctions.pop_back();
+      --i; --e;
+    }
+}
+
+/// removeOneAbstractEdgeTo - Remove one edge associated with a null callsite
+/// from this node to the specified callee function.
+void CallGraphNode::removeOneAbstractEdgeTo(CallGraphNode *Callee) {
+  for (CalledFunctionsVector::iterator I = CalledFunctions.begin(); ; ++I) {
+    assert(I != CalledFunctions.end() && "Cannot find callee to remove!");
+    CallRecord &CR = *I;
+    if (CR.second == Callee && !CR.first.getInstruction()) {
+      CalledFunctions.erase(I);
+      return;
+    }
+  }
+}
+
+/// replaceCallSite - Make the edge in the node for Old CallSite be for
+/// New CallSite instead.  Note that this method takes linear time, so it
+/// should be used sparingly.
+void CallGraphNode::replaceCallSite(CallSite Old, CallSite New) {
+  for (CalledFunctionsVector::iterator I = CalledFunctions.begin(); ; ++I) {
+    assert(I != CalledFunctions.end() && "Cannot find callsite to replace!");
+    if (I->first == Old) {
+      I->first = New;
+      return;
+    }
+  }
+}
+
+// Enuse that users of CallGraph.h also link with this file
+DEFINING_FILE_FOR(CallGraph)
diff --git a/lib/Analysis/IPA/CallGraphSCCPass.cpp b/lib/Analysis/IPA/CallGraphSCCPass.cpp
new file mode 100644
index 0000000..3880d0a
--- /dev/null
+++ b/lib/Analysis/IPA/CallGraphSCCPass.cpp
@@ -0,0 +1,207 @@
+//===- CallGraphSCCPass.cpp - Pass that operates BU on call graph ---------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This file implements the CallGraphSCCPass class, which is used for passes
+// which are implemented as bottom-up traversals on the call graph.  Because
+// there may be cycles in the call graph, passes of this type operate on the
+// call-graph in SCC order: that is, they process function bottom-up, except for
+// recursive functions, which they process all at once.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/CallGraphSCCPass.h"
+#include "llvm/Analysis/CallGraph.h"
+#include "llvm/ADT/SCCIterator.h"
+#include "llvm/PassManagers.h"
+#include "llvm/Function.h"
+using namespace llvm;
+
+//===----------------------------------------------------------------------===//
+// CGPassManager
+//
+/// CGPassManager manages FPPassManagers and CalLGraphSCCPasses.
+
+namespace {
+
+class CGPassManager : public ModulePass, public PMDataManager {
+
+public:
+  static char ID;
+  explicit CGPassManager(int Depth) 
+    : ModulePass(&ID), PMDataManager(Depth) { }
+
+  /// run - Execute all of the passes scheduled for execution.  Keep track of
+  /// whether any of the passes modifies the module, and if so, return true.
+  bool runOnModule(Module &M);
+
+  bool doInitialization(CallGraph &CG);
+  bool doFinalization(CallGraph &CG);
+
+  /// Pass Manager itself does not invalidate any analysis info.
+  void getAnalysisUsage(AnalysisUsage &Info) const {
+    // CGPassManager walks SCC and it needs CallGraph.
+    Info.addRequired<CallGraph>();
+    Info.setPreservesAll();
+  }
+
+  virtual const char *getPassName() const {
+    return "CallGraph Pass Manager";
+  }
+
+  // Print passes managed by this manager
+  void dumpPassStructure(unsigned Offset) {
+    llvm::cerr << std::string(Offset*2, ' ') << "Call Graph SCC Pass Manager\n";
+    for (unsigned Index = 0; Index < getNumContainedPasses(); ++Index) {
+      Pass *P = getContainedPass(Index);
+      P->dumpPassStructure(Offset + 1);
+      dumpLastUses(P, Offset+1);
+    }
+  }
+
+  Pass *getContainedPass(unsigned N) {
+    assert ( N < PassVector.size() && "Pass number out of range!");
+    Pass *FP = static_cast<Pass *>(PassVector[N]);
+    return FP;
+  }
+
+  virtual PassManagerType getPassManagerType() const { 
+    return PMT_CallGraphPassManager; 
+  }
+};
+
+}
+
+char CGPassManager::ID = 0;
+/// run - Execute all of the passes scheduled for execution.  Keep track of
+/// whether any of the passes modifies the module, and if so, return true.
+bool CGPassManager::runOnModule(Module &M) {
+  CallGraph &CG = getAnalysis<CallGraph>();
+  bool Changed = doInitialization(CG);
+
+  // Walk SCC
+  for (scc_iterator<CallGraph*> I = scc_begin(&CG), E = scc_end(&CG);
+       I != E; ++I) {
+
+    // Run all passes on current SCC
+    for (unsigned Index = 0; Index < getNumContainedPasses(); ++Index) {
+      Pass *P = getContainedPass(Index);
+
+      dumpPassInfo(P, EXECUTION_MSG, ON_CG_MSG, "");
+      dumpRequiredSet(P);
+
+      initializeAnalysisImpl(P);
+
+      StartPassTimer(P);
+      if (CallGraphSCCPass *CGSP = dynamic_cast<CallGraphSCCPass *>(P))
+        Changed |= CGSP->runOnSCC(*I);   // TODO : What if CG is changed ?
+      else {
+        FPPassManager *FPP = dynamic_cast<FPPassManager *>(P);
+        assert (FPP && "Invalid CGPassManager member");
+
+        // Run pass P on all functions current SCC
+        std::vector<CallGraphNode*> &SCC = *I;
+        for (unsigned i = 0, e = SCC.size(); i != e; ++i) {
+          Function *F = SCC[i]->getFunction();
+          if (F) {
+            dumpPassInfo(P, EXECUTION_MSG, ON_FUNCTION_MSG, F->getNameStart());
+            Changed |= FPP->runOnFunction(*F);
+          }
+        }
+      }
+      StopPassTimer(P);
+
+      if (Changed)
+        dumpPassInfo(P, MODIFICATION_MSG, ON_CG_MSG, "");
+      dumpPreservedSet(P);
+
+      verifyPreservedAnalysis(P);      
+      removeNotPreservedAnalysis(P);
+      recordAvailableAnalysis(P);
+      removeDeadPasses(P, "", ON_CG_MSG);
+    }
+  }
+  Changed |= doFinalization(CG);
+  return Changed;
+}
+
+/// Initialize CG
+bool CGPassManager::doInitialization(CallGraph &CG) {
+  bool Changed = false;
+  for (unsigned Index = 0; Index < getNumContainedPasses(); ++Index) {  
+    Pass *P = getContainedPass(Index);
+    if (CallGraphSCCPass *CGSP = dynamic_cast<CallGraphSCCPass *>(P)) {
+      Changed |= CGSP->doInitialization(CG);
+    } else {
+      FPPassManager *FP = dynamic_cast<FPPassManager *>(P);
+      assert (FP && "Invalid CGPassManager member");
+      Changed |= FP->doInitialization(CG.getModule());
+    }
+  }
+  return Changed;
+}
+
+/// Finalize CG
+bool CGPassManager::doFinalization(CallGraph &CG) {
+  bool Changed = false;
+  for (unsigned Index = 0; Index < getNumContainedPasses(); ++Index) {  
+    Pass *P = getContainedPass(Index);
+    if (CallGraphSCCPass *CGSP = dynamic_cast<CallGraphSCCPass *>(P)) {
+      Changed |= CGSP->doFinalization(CG);
+    } else {
+      FPPassManager *FP = dynamic_cast<FPPassManager *>(P);
+      assert (FP && "Invalid CGPassManager member");
+      Changed |= FP->doFinalization(CG.getModule());
+    }
+  }
+  return Changed;
+}
+
+/// Assign pass manager to manage this pass.
+void CallGraphSCCPass::assignPassManager(PMStack &PMS,
+                                         PassManagerType PreferredType) {
+  // Find CGPassManager 
+  while (!PMS.empty() &&
+         PMS.top()->getPassManagerType() > PMT_CallGraphPassManager)
+    PMS.pop();
+
+  assert (!PMS.empty() && "Unable to handle Call Graph Pass");
+  CGPassManager *CGP = dynamic_cast<CGPassManager *>(PMS.top());
+
+  // Create new Call Graph SCC Pass Manager if it does not exist. 
+  if (!CGP) {
+
+    assert (!PMS.empty() && "Unable to create Call Graph Pass Manager");
+    PMDataManager *PMD = PMS.top();
+
+    // [1] Create new Call Graph Pass Manager
+    CGP = new CGPassManager(PMD->getDepth() + 1);
+
+    // [2] Set up new manager's top level manager
+    PMTopLevelManager *TPM = PMD->getTopLevelManager();
+    TPM->addIndirectPassManager(CGP);
+
+    // [3] Assign manager to manage this new manager. This may create
+    // and push new managers into PMS
+    Pass *P = dynamic_cast<Pass *>(CGP);
+    TPM->schedulePass(P);
+
+    // [4] Push new manager into PMS
+    PMS.push(CGP);
+  }
+
+  CGP->add(this);
+}
+
+/// getAnalysisUsage - For this class, we declare that we require and preserve
+/// the call graph.  If the derived class implements this method, it should
+/// always explicitly call the implementation here.
+void CallGraphSCCPass::getAnalysisUsage(AnalysisUsage &AU) const {
+  AU.addRequired<CallGraph>();
+  AU.addPreserved<CallGraph>();
+}
diff --git a/lib/Analysis/IPA/FindUsedTypes.cpp b/lib/Analysis/IPA/FindUsedTypes.cpp
new file mode 100644
index 0000000..920ee37
--- /dev/null
+++ b/lib/Analysis/IPA/FindUsedTypes.cpp
@@ -0,0 +1,104 @@
+//===- FindUsedTypes.cpp - Find all Types used by a module ----------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This pass is used to seek out all of the types in use by the program.  Note
+// that this analysis explicitly does not include types only used by the symbol
+// table.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Analysis/FindUsedTypes.h"
+#include "llvm/Constants.h"
+#include "llvm/DerivedTypes.h"
+#include "llvm/Module.h"
+#include "llvm/Assembly/Writer.h"
+#include "llvm/Support/InstIterator.h"
+#include "llvm/Support/raw_ostream.h"
+using namespace llvm;
+
+char FindUsedTypes::ID = 0;
+static RegisterPass<FindUsedTypes>
+X("print-used-types", "Find Used Types", false, true);
+
+// IncorporateType - Incorporate one type and all of its subtypes into the
+// collection of used types.
+//
+void FindUsedTypes::IncorporateType(const Type *Ty) {
+  // If ty doesn't already exist in the used types map, add it now, otherwise
+  // return.
+  if (!UsedTypes.insert(Ty).second) return;  // Already contain Ty.
+
+  // Make sure to add any types this type references now.
+  //
+  for (Type::subtype_iterator I = Ty->subtype_begin(), E = Ty->subtype_end();
+       I != E; ++I)
+    IncorporateType(*I);
+}
+
+void FindUsedTypes::IncorporateValue(const Value *V) {
+  IncorporateType(V->getType());
+
+  // If this is a constant, it could be using other types...
+  if (const Constant *C = dyn_cast<Constant>(V)) {
+    if (!isa<GlobalValue>(C))
+      for (User::const_op_iterator OI = C->op_begin(), OE = C->op_end();
+           OI != OE; ++OI)
+        IncorporateValue(*OI);
+  }
+}
+
+
+// run - This incorporates all types used by the specified module
+//
+bool FindUsedTypes::runOnModule(Module &m) {
+  UsedTypes.clear();  // reset if run multiple times...
+
+  // Loop over global variables, incorporating their types
+  for (Module::const_global_iterator I = m.global_begin(), E = m.global_end();
+       I != E; ++I) {
+    IncorporateType(I->getType());
+    if (I->hasInitializer())
+      IncorporateValue(I->getInitializer());
+  }
+
+  for (Module::iterator MI = m.begin(), ME = m.end(); MI != ME; ++MI) {
+    IncorporateType(MI->getType());
+    const Function &F = *MI;
+
+    // Loop over all of the instructions in the function, adding their return
+    // type as well as the types of their operands.
+    //
+    for (const_inst_iterator II = inst_begin(F), IE = inst_end(F);
+         II != IE; ++II) {
+      const Instruction &I = *II;
+
+      IncorporateType(I.getType());  // Incorporate the type of the instruction
+      for (User::const_op_iterator OI = I.op_begin(), OE = I.op_end();
+           OI != OE; ++OI)
+        IncorporateValue(*OI);  // Insert inst operand types as well
+    }
+  }
+
+  return false;
+}
+
+// Print the types found in the module.  If the optional Module parameter is
+// passed in, then the types are printed symbolically if possible, using the
+// symbol table from the module.
+//
+void FindUsedTypes::print(std::ostream &OS, const Module *M) const {
+  raw_os_ostream RO(OS);
+  RO << "Types in use by this module:\n";
+  for (std::set<const Type *>::const_iterator I = UsedTypes.begin(),
+       E = UsedTypes.end(); I != E; ++I) {
+    RO << "   ";
+    WriteTypeSymbolic(RO, *I, M);
+    RO << '\n';
+  }
+}
diff --git a/lib/Analysis/IPA/GlobalsModRef.cpp b/lib/Analysis/IPA/GlobalsModRef.cpp
new file mode 100644
index 0000000..2e9884a
--- /dev/null
+++ b/lib/Analysis/IPA/GlobalsModRef.cpp
@@ -0,0 +1,567 @@
+//===- GlobalsModRef.cpp - Simple Mod/Ref Analysis for Globals ------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This simple pass provides alias and mod/ref information for global values
+// that do not have their address taken, and keeps track of whether functions
+// read or write memory (are "pure").  For this simple (but very common) case,
+// we can provide pretty accurate and useful information.
+//
+//===----------------------------------------------------------------------===//
+
+#define DEBUG_TYPE "globalsmodref-aa"
+#include "llvm/Analysis/Passes.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+#include "llvm/Instructions.h"
+#include "llvm/Constants.h"
+#include "llvm/DerivedTypes.h"
+#include "llvm/Analysis/AliasAnalysis.h"
+#include "llvm/Analysis/CallGraph.h"
+#include "llvm/Support/Compiler.h"
+#include "llvm/Support/CommandLine.h"
+#include "llvm/Support/InstIterator.h"
+#include "llvm/ADT/Statistic.h"
+#include "llvm/ADT/SCCIterator.h"
+#include <set>
+using namespace llvm;
+
+STATISTIC(NumNonAddrTakenGlobalVars,
+          "Number of global vars without address taken");
+STATISTIC(NumNonAddrTakenFunctions,"Number of functions without address taken");
+STATISTIC(NumNoMemFunctions, "Number of functions that do not access memory");
+STATISTIC(NumReadMemFunctions, "Number of functions that only read memory");
+STATISTIC(NumIndirectGlobalVars, "Number of indirect global objects");
+
+namespace {
+  /// FunctionRecord - One instance of this structure is stored for every
+  /// function in the program.  Later, the entries for these functions are
+  /// removed if the function is found to call an external function (in which
+  /// case we know nothing about it.
+  struct VISIBILITY_HIDDEN FunctionRecord {
+    /// GlobalInfo - Maintain mod/ref info for all of the globals without
+    /// addresses taken that are read or written (transitively) by this
+    /// function.
+    std::map<GlobalValue*, unsigned> GlobalInfo;
+
+    /// MayReadAnyGlobal - May read global variables, but it is not known which.
+    bool MayReadAnyGlobal;
+
+    unsigned getInfoForGlobal(GlobalValue *GV) const {
+      unsigned Effect = MayReadAnyGlobal ? AliasAnalysis::Ref : 0;
+      std::map<GlobalValue*, unsigned>::const_iterator I = GlobalInfo.find(GV);
+      if (I != GlobalInfo.end())
+        Effect |= I->second;
+      return Effect;
+    }
+
+    /// FunctionEffect - Capture whether or not this function reads or writes to
+    /// ANY memory.  If not, we can do a lot of aggressive analysis on it.
+    unsigned FunctionEffect;
+
+    FunctionRecord() : MayReadAnyGlobal (false), FunctionEffect(0) {}
+  };
+
+  /// GlobalsModRef - The actual analysis pass.
+  class VISIBILITY_HIDDEN GlobalsModRef
+      : public ModulePass, public AliasAnalysis {
+    /// NonAddressTakenGlobals - The globals that do not have their addresses
+    /// taken.
+    std::set<GlobalValue*> NonAddressTakenGlobals;
+
+    /// IndirectGlobals - The memory pointed to by this global is known to be
+    /// 'owned' by the global.
+    std::set<GlobalValue*> IndirectGlobals;
+
+    /// AllocsForIndirectGlobals - If an instruction allocates memory for an
+    /// indirect global, this map indicates which one.
+    std::map<Value*, GlobalValue*> AllocsForIndirectGlobals;
+
+    /// FunctionInfo - For each function, keep track of what globals are
+    /// modified or read.
+    std::map<Function*, FunctionRecord> FunctionInfo;
+
+  public:
+    static char ID;
+    GlobalsModRef() : ModulePass(&ID) {}
+
+    bool runOnModule(Module &M) {
+      InitializeAliasAnalysis(this);                 // set up super class
+      AnalyzeGlobals(M);                          // find non-addr taken globals
+      AnalyzeCallGraph(getAnalysis<CallGraph>(), M); // Propagate on CG
+      return false;
+    }
+
+    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
+      AliasAnalysis::getAnalysisUsage(AU);
+      AU.addRequired<CallGraph>();
+      AU.setPreservesAll();                         // Does not transform code
+    }
+
+    //------------------------------------------------
+    // Implement the AliasAnalysis API
+    //
+    AliasResult alias(const Value *V1, unsigned V1Size,
+                      const Value *V2, unsigned V2Size);
+    ModRefResult getModRefInfo(CallSite CS, Value *P, unsigned Size);
+    ModRefResult getModRefInfo(CallSite CS1, CallSite CS2) {
+      return AliasAnalysis::getModRefInfo(CS1,CS2);
+    }
+    bool hasNoModRefInfoForCalls() const { return false; }
+
+    /// getModRefBehavior - Return the behavior of the specified function if
+    /// called from the specified call site.  The call site may be null in which
+    /// case the most generic behavior of this function should be returned.
+    ModRefBehavior getModRefBehavior(Function *F,
+                                         std::vector<PointerAccessInfo> *Info) {
+      if (FunctionRecord *FR = getFunctionInfo(F)) {
+        if (FR->FunctionEffect == 0)
+          return DoesNotAccessMemory;
+        else if ((FR->FunctionEffect & Mod) == 0)
+          return OnlyReadsMemory;
+      }
+      return AliasAnalysis::getModRefBehavior(F, Info);
+    }
+    
+    /// getModRefBehavior - Return the behavior of the specified function if
+    /// called from the specified call site.  The call site may be null in which
+    /// case the most generic behavior of this function should be returned.
+    ModRefBehavior getModRefBehavior(CallSite CS,
+                                         std::vector<PointerAccessInfo> *Info) {
+      Function* F = CS.getCalledFunction();
+      if (!F) return AliasAnalysis::getModRefBehavior(CS, Info);
+      if (FunctionRecord *FR = getFunctionInfo(F)) {
+        if (FR->FunctionEffect == 0)
+          return DoesNotAccessMemory;
+        else if ((FR->FunctionEffect & Mod) == 0)
+          return OnlyReadsMemory;
+      }
+      return AliasAnalysis::getModRefBehavior(CS, Info);
+    }
+
+    virtual void deleteValue(Value *V);
+    virtual void copyValue(Value *From, Value *To);
+
+  private:
+    /// getFunctionInfo - Return the function info for the function, or null if
+    /// we don't have anything useful to say about it.
+    FunctionRecord *getFunctionInfo(Function *F) {
+      std::map<Function*, FunctionRecord>::iterator I = FunctionInfo.find(F);
+      if (I != FunctionInfo.end())
+        return &I->second;
+      return 0;
+    }
+
+    void AnalyzeGlobals(Module &M);
+    void AnalyzeCallGraph(CallGraph &CG, Module &M);
+    bool AnalyzeUsesOfPointer(Value *V, std::vector<Function*> &Readers,
+                              std::vector<Function*> &Writers,
+                              GlobalValue *OkayStoreDest = 0);
+    bool AnalyzeIndirectGlobalMemory(GlobalValue *GV);
+  };
+}
+
+char GlobalsModRef::ID = 0;
+static RegisterPass<GlobalsModRef>
+X("globalsmodref-aa", "Simple mod/ref analysis for globals", false, true);
+static RegisterAnalysisGroup<AliasAnalysis> Y(X);
+
+Pass *llvm::createGlobalsModRefPass() { return new GlobalsModRef(); }
+
+/// AnalyzeGlobals - Scan through the users of all of the internal
+/// GlobalValue's in the program.  If none of them have their "address taken"
+/// (really, their address passed to something nontrivial), record this fact,
+/// and record the functions that they are used directly in.
+void GlobalsModRef::AnalyzeGlobals(Module &M) {
+  std::vector<Function*> Readers, Writers;
+  for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I)
+    if (I->hasLocalLinkage()) {
+      if (!AnalyzeUsesOfPointer(I, Readers, Writers)) {
+        // Remember that we are tracking this global.
+        NonAddressTakenGlobals.insert(I);
+        ++NumNonAddrTakenFunctions;
+      }
+      Readers.clear(); Writers.clear();
+    }
+
+  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
+       I != E; ++I)
+    if (I->hasLocalLinkage()) {
+      if (!AnalyzeUsesOfPointer(I, Readers, Writers)) {
+        // Remember that we are tracking this global, and the mod/ref fns
+        NonAddressTakenGlobals.insert(I);
+
+        for (unsigned i = 0, e = Readers.size(); i != e; ++i)
+          FunctionInfo[Readers[i]].GlobalInfo[I] |= Ref;
+
+        if (!I->isConstant())  // No need to keep track of writers to constants
+          for (unsigned i = 0, e = Writers.size(); i != e; ++i)
+            FunctionInfo[Writers[i]].GlobalInfo[I] |= Mod;
+        ++NumNonAddrTakenGlobalVars;
+
+        // If this global holds a pointer type, see if it is an indirect global.
+        if (isa<PointerType>(I->getType()->getElementType()) &&
+            AnalyzeIndirectGlobalMemory(I))
+          ++NumIndirectGlobalVars;
+      }
+      Readers.clear(); Writers.clear();
+    }
+}
+
+/// AnalyzeUsesOfPointer - Look at all of the users of the specified pointer.
+/// If this is used by anything complex (i.e., the address escapes), return
+/// true.  Also, while we are at it, keep track of those functions that read and
+/// write to the value.
+///
+/// If OkayStoreDest is non-null, stores into this global are allowed.
+bool GlobalsModRef::AnalyzeUsesOfPointer(Value *V,
+                                         std::vector<Function*> &Readers,
+                                         std::vector<Function*> &Writers,
+                                         GlobalValue *OkayStoreDest) {
+  if (!isa<PointerType>(V->getType())) return true;
+
+  for (Value::use_iterator UI = V->use_begin(), E = V->use_end(); UI != E; ++UI)
+    if (LoadInst *LI = dyn_cast<LoadInst>(*UI)) {
+      Readers.push_back(LI->getParent()->getParent());
+    } else if (StoreInst *SI = dyn_cast<StoreInst>(*UI)) {
+      if (V == SI->getOperand(1)) {
+        Writers.push_back(SI->getParent()->getParent());
+      } else if (SI->getOperand(1) != OkayStoreDest) {
+        return true;  // Storing the pointer
+      }
+    } else if (GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(*UI)) {
+      if (AnalyzeUsesOfPointer(GEP, Readers, Writers)) return true;
+    } else if (CallInst *CI = dyn_cast<CallInst>(*UI)) {
+      // Make sure that this is just the function being called, not that it is
+      // passing into the function.
+      for (unsigned i = 1, e = CI->getNumOperands(); i != e; ++i)
+        if (CI->getOperand(i) == V) return true;
+    } else if (InvokeInst *II = dyn_cast<InvokeInst>(*UI)) {
+      // Make sure that this is just the function being called, not that it is
+      // passing into the function.
+      for (unsigned i = 3, e = II->getNumOperands(); i != e; ++i)
+        if (II->getOperand(i) == V) return true;
+    } else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(*UI)) {
+      if (CE->getOpcode() == Instruction::GetElementPtr ||
+          CE->getOpcode() == Instruction::BitCast) {
+        if (AnalyzeUsesOfPointer(CE, Readers, Writers))
+          return true;
+      } else {
+        return true;
+      }
+    } else if (ICmpInst *ICI = dyn_cast<ICmpInst>(*UI)) {
+      if (!isa<ConstantPointerNull>(ICI->getOperand(1)))
+        return true;  // Allow comparison against null.
+    } else if (FreeInst *F = dyn_cast<FreeInst>(*UI)) {
+      Writers.push_back(F->getParent()->getParent());
+    } else {
+      return true;
+    }
+  return false;
+}
+
+/// AnalyzeIndirectGlobalMemory - We found an non-address-taken global variable
+/// which holds a pointer type.  See if the global always points to non-aliased
+/// heap memory: that is, all initializers of the globals are allocations, and
+/// those allocations have no use other than initialization of the global.
+/// Further, all loads out of GV must directly use the memory, not store the
+/// pointer somewhere.  If this is true, we consider the memory pointed to by
+/// GV to be owned by GV and can disambiguate other pointers from it.
+bool GlobalsModRef::AnalyzeIndirectGlobalMemory(GlobalValue *GV) {
+  // Keep track of values related to the allocation of the memory, f.e. the
+  // value produced by the malloc call and any casts.
+  std::vector<Value*> AllocRelatedValues;
+
+  // Walk the user list of the global.  If we find anything other than a direct
+  // load or store, bail out.
+  for (Value::use_iterator I = GV->use_begin(), E = GV->use_end(); I != E; ++I){
+    if (LoadInst *LI = dyn_cast<LoadInst>(*I)) {
+      // The pointer loaded from the global can only be used in simple ways:
+      // we allow addressing of it and loading storing to it.  We do *not* allow
+      // storing the loaded pointer somewhere else or passing to a function.
+      std::vector<Function*> ReadersWriters;
+      if (AnalyzeUsesOfPointer(LI, ReadersWriters, ReadersWriters))
+        return false;  // Loaded pointer escapes.
+      // TODO: Could try some IP mod/ref of the loaded pointer.
+    } else if (StoreInst *SI = dyn_cast<StoreInst>(*I)) {
+      // Storing the global itself.
+      if (SI->getOperand(0) == GV) return false;
+
+      // If storing the null pointer, ignore it.
+      if (isa<ConstantPointerNull>(SI->getOperand(0)))
+        continue;
+
+      // Check the value being stored.
+      Value *Ptr = SI->getOperand(0)->getUnderlyingObject();
+
+      if (isa<MallocInst>(Ptr)) {
+        // Okay, easy case.
+      } else if (CallInst *CI = dyn_cast<CallInst>(Ptr)) {
+        Function *F = CI->getCalledFunction();
+        if (!F || !F->isDeclaration()) return false;     // Too hard to analyze.
+        if (F->getName() != "calloc") return false;   // Not calloc.
+      } else {
+        return false;  // Too hard to analyze.
+      }
+
+      // Analyze all uses of the allocation.  If any of them are used in a
+      // non-simple way (e.g. stored to another global) bail out.
+      std::vector<Function*> ReadersWriters;
+      if (AnalyzeUsesOfPointer(Ptr, ReadersWriters, ReadersWriters, GV))
+        return false;  // Loaded pointer escapes.
+
+      // Remember that this allocation is related to the indirect global.
+      AllocRelatedValues.push_back(Ptr);
+    } else {
+      // Something complex, bail out.
+      return false;
+    }
+  }
+
+  // Okay, this is an indirect global.  Remember all of the allocations for
+  // this global in AllocsForIndirectGlobals.
+  while (!AllocRelatedValues.empty()) {
+    AllocsForIndirectGlobals[AllocRelatedValues.back()] = GV;
+    AllocRelatedValues.pop_back();
+  }
+  IndirectGlobals.insert(GV);
+  return true;
+}
+
+/// AnalyzeCallGraph - At this point, we know the functions where globals are
+/// immediately stored to and read from.  Propagate this information up the call
+/// graph to all callers and compute the mod/ref info for all memory for each
+/// function.
+void GlobalsModRef::AnalyzeCallGraph(CallGraph &CG, Module &M) {
+  // We do a bottom-up SCC traversal of the call graph.  In other words, we
+  // visit all callees before callers (leaf-first).
+  for (scc_iterator<CallGraph*> I = scc_begin(&CG), E = scc_end(&CG); I != E;
+       ++I) {
+    std::vector<CallGraphNode *> &SCC = *I;
+    assert(!SCC.empty() && "SCC with no functions?");
+
+    if (!SCC[0]->getFunction()) {
+      // Calls externally - can't say anything useful.  Remove any existing
+      // function records (may have been created when scanning globals).
+      for (unsigned i = 0, e = SCC.size(); i != e; ++i)
+        FunctionInfo.erase(SCC[i]->getFunction());
+      continue;
+    }
+
+    FunctionRecord &FR = FunctionInfo[SCC[0]->getFunction()];
+
+    bool KnowNothing = false;
+    unsigned FunctionEffect = 0;
+
+    // Collect the mod/ref properties due to called functions.  We only compute
+    // one mod-ref set.
+    for (unsigned i = 0, e = SCC.size(); i != e && !KnowNothing; ++i) {
+      Function *F = SCC[i]->getFunction();
+      if (!F) {
+        KnowNothing = true;
+        break;
+      }
+
+      if (F->isDeclaration()) {
+        // Try to get mod/ref behaviour from function attributes.
+        if (F->doesNotAccessMemory()) {
+          // Can't do better than that!
+        } else if (F->onlyReadsMemory()) {
+          FunctionEffect |= Ref;
+          if (!F->isIntrinsic())
+            // This function might call back into the module and read a global -
+            // consider every global as possibly being read by this function.
+            FR.MayReadAnyGlobal = true;
+        } else {
+          FunctionEffect |= ModRef;
+          // Can't say anything useful unless it's an intrinsic - they don't
+          // read or write global variables of the kind considered here.
+          KnowNothing = !F->isIntrinsic();
+        }
+        continue;
+      }
+
+      for (CallGraphNode::iterator CI = SCC[i]->begin(), E = SCC[i]->end();
+           CI != E && !KnowNothing; ++CI)
+        if (Function *Callee = CI->second->getFunction()) {
+          if (FunctionRecord *CalleeFR = getFunctionInfo(Callee)) {
+            // Propagate function effect up.
+            FunctionEffect |= CalleeFR->FunctionEffect;
+
+            // Incorporate callee's effects on globals into our info.
+            for (std::map<GlobalValue*, unsigned>::iterator GI =
+                   CalleeFR->GlobalInfo.begin(), E = CalleeFR->GlobalInfo.end();
+                 GI != E; ++GI)
+              FR.GlobalInfo[GI->first] |= GI->second;
+            FR.MayReadAnyGlobal |= CalleeFR->MayReadAnyGlobal;
+          } else {
+            // Can't say anything about it.  However, if it is inside our SCC,
+            // then nothing needs to be done.
+            CallGraphNode *CalleeNode = CG[Callee];
+            if (std::find(SCC.begin(), SCC.end(), CalleeNode) == SCC.end())
+              KnowNothing = true;
+          }
+        } else {
+          KnowNothing = true;
+        }
+    }
+
+    // If we can't say anything useful about this SCC, remove all SCC functions
+    // from the FunctionInfo map.
+    if (KnowNothing) {
+      for (unsigned i = 0, e = SCC.size(); i != e; ++i)
+        FunctionInfo.erase(SCC[i]->getFunction());
+      continue;
+    }
+
+    // Scan the function bodies for explicit loads or stores.
+    for (unsigned i = 0, e = SCC.size(); i != e && FunctionEffect != ModRef;++i)
+      for (inst_iterator II = inst_begin(SCC[i]->getFunction()),
+             E = inst_end(SCC[i]->getFunction());
+           II != E && FunctionEffect != ModRef; ++II)
+        if (isa<LoadInst>(*II)) {
+          FunctionEffect |= Ref;
+          if (cast<LoadInst>(*II).isVolatile())
+            // Volatile loads may have side-effects, so mark them as writing
+            // memory (for example, a flag inside the processor).
+            FunctionEffect |= Mod;
+        } else if (isa<StoreInst>(*II)) {
+          FunctionEffect |= Mod;
+          if (cast<StoreInst>(*II).isVolatile())
+            // Treat volatile stores as reading memory somewhere.
+            FunctionEffect |= Ref;
+        } else if (isa<MallocInst>(*II) || isa<FreeInst>(*II)) {
+          FunctionEffect |= ModRef;
+        }
+
+    if ((FunctionEffect & Mod) == 0)
+      ++NumReadMemFunctions;
+    if (FunctionEffect == 0)
+      ++NumNoMemFunctions;
+    FR.FunctionEffect = FunctionEffect;
+
+    // Finally, now that we know the full effect on this SCC, clone the
+    // information to each function in the SCC.
+    for (unsigned i = 1, e = SCC.size(); i != e; ++i)
+      FunctionInfo[SCC[i]->getFunction()] = FR;
+  }
+}
+
+
+
+/// alias - If one of the pointers is to a global that we are tracking, and the
+/// other is some random pointer, we know there cannot be an alias, because the
+/// address of the global isn't taken.
+AliasAnalysis::AliasResult
+GlobalsModRef::alias(const Value *V1, unsigned V1Size,
+                     const Value *V2, unsigned V2Size) {
+  // Get the base object these pointers point to.
+  Value *UV1 = const_cast<Value*>(V1->getUnderlyingObject());
+  Value *UV2 = const_cast<Value*>(V2->getUnderlyingObject());
+
+  // If either of the underlying values is a global, they may be non-addr-taken
+  // globals, which we can answer queries about.
+  GlobalValue *GV1 = dyn_cast<GlobalValue>(UV1);
+  GlobalValue *GV2 = dyn_cast<GlobalValue>(UV2);
+  if (GV1 || GV2) {
+    // If the global's address is taken, pretend we don't know it's a pointer to
+    // the global.
+    if (GV1 && !NonAddressTakenGlobals.count(GV1)) GV1 = 0;
+    if (GV2 && !NonAddressTakenGlobals.count(GV2)) GV2 = 0;
+
+    // If the the two pointers are derived from two different non-addr-taken
+    // globals, or if one is and the other isn't, we know these can't alias.
+    if ((GV1 || GV2) && GV1 != GV2)
+      return NoAlias;
+
+    // Otherwise if they are both derived from the same addr-taken global, we
+    // can't know the two accesses don't overlap.
+  }
+
+  // These pointers may be based on the memory owned by an indirect global.  If
+  // so, we may be able to handle this.  First check to see if the base pointer
+  // is a direct load from an indirect global.
+  GV1 = GV2 = 0;
+  if (LoadInst *LI = dyn_cast<LoadInst>(UV1))
+    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(LI->getOperand(0)))
+      if (IndirectGlobals.count(GV))
+        GV1 = GV;
+  if (LoadInst *LI = dyn_cast<LoadInst>(UV2))
+    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(LI->getOperand(0)))
+      if (IndirectGlobals.count(GV))
+        GV2 = GV;
+
+  // These pointers may also be from an allocation for the indirect global.  If
+  // so, also handle them.
+  if (AllocsForIndirectGlobals.count(UV1))
+    GV1 = AllocsForIndirectGlobals[UV1];
+  if (AllocsForIndirectGlobals.count(UV2))
+    GV2 = AllocsForIndirectGlobals[UV2];
+
+  // Now that we know whether the two pointers are related to indirect globals,
+  // use this to disambiguate the pointers.  If either pointer is based on an
+  // indirect global and if they are not both based on the same indirect global,
+  // they cannot alias.
+  if ((GV1 || GV2) && GV1 != GV2)
+    return NoAlias;
+
+  return AliasAnalysis::alias(V1, V1Size, V2, V2Size);
+}
+
+AliasAnalysis::ModRefResult
+GlobalsModRef::getModRefInfo(CallSite CS, Value *P, unsigned Size) {
+  unsigned Known = ModRef;
+
+  // If we are asking for mod/ref info of a direct call with a pointer to a
+  // global we are tracking, return information if we have it.
+  if (GlobalValue *GV = dyn_cast<GlobalValue>(P->getUnderlyingObject()))
+    if (GV->hasLocalLinkage())
+      if (Function *F = CS.getCalledFunction())
+        if (NonAddressTakenGlobals.count(GV))
+          if (FunctionRecord *FR = getFunctionInfo(F))
+            Known = FR->getInfoForGlobal(GV);
+
+  if (Known == NoModRef)
+    return NoModRef; // No need to query other mod/ref analyses
+  return ModRefResult(Known & AliasAnalysis::getModRefInfo(CS, P, Size));
+}
+
+
+//===----------------------------------------------------------------------===//
+// Methods to update the analysis as a result of the client transformation.
+//
+void GlobalsModRef::deleteValue(Value *V) {
+  if (GlobalValue *GV = dyn_cast<GlobalValue>(V)) {
+    if (NonAddressTakenGlobals.erase(GV)) {
+      // This global might be an indirect global.  If so, remove it and remove
+      // any AllocRelatedValues for it.
+      if (IndirectGlobals.erase(GV)) {
+        // Remove any entries in AllocsForIndirectGlobals for this global.
+        for (std::map<Value*, GlobalValue*>::iterator
+             I = AllocsForIndirectGlobals.begin(),
+             E = AllocsForIndirectGlobals.end(); I != E; ) {
+          if (I->second == GV) {
+            AllocsForIndirectGlobals.erase(I++);
+          } else {
+            ++I;
+          }
+        }
+      }
+    }
+  }
+
+  // Otherwise, if this is an allocation related to an indirect global, remove
+  // it.
+  AllocsForIndirectGlobals.erase(V);
+
+  AliasAnalysis::deleteValue(V);
+}
+
+void GlobalsModRef::copyValue(Value *From, Value *To) {
+  AliasAnalysis::copyValue(From, To);
+}
diff --git a/lib/Analysis/IPA/Makefile b/lib/Analysis/IPA/Makefile
new file mode 100644
index 0000000..adacb16
--- /dev/null
+++ b/lib/Analysis/IPA/Makefile
@@ -0,0 +1,14 @@
+##===- lib/Analysis/IPA/Makefile ---------------------------*- Makefile -*-===##
+#
+#                     The LLVM Compiler Infrastructure
+#
+# This file is distributed under the University of Illinois Open Source
+# License. See LICENSE.TXT for details.
+#
+##===----------------------------------------------------------------------===##
+
+LEVEL = ../../..
+LIBRARYNAME = LLVMipa
+BUILD_ARCHIVE = 1
+include $(LEVEL)/Makefile.common
+