[PM] Port memdep to the new pass manager.

[chandlerc]

Summary

This is a fairly straightforward port to the new pass manager with one
exception. It removes a very questionable use of releaseMemory() in
the old pass to invalidate its caches between runs on a function.
I don't think this is really guaranteed to be safe. I've just used the
more direct port to the new PM to address this by nuking the results
object each time the pass runs. While this could cause some minor malloc
traffic increase, I don't expect the compile time performance hit to be
noticable, and it makes the correctness and other aspects of the pass
much easier to reason about.

That said I'm sending it out for review so that Mehdi can help me confirm that
there is no obvious and significant compile time regression here.

There is sadly very limited testing at this point as there are only two
tests of memdep, and both rely on GVN. I'll be porting GVN next and that
will exercise this heavily though.

[bruno]

Hi Chandler,

This is great!
I could not see any compile time regressions while running -O3 against the test-suite. But got more improvements/regressions results in -O2:

Compile-time O2

Performance Regressions - Compile Time Δ
External/SPEC/CFP2006/470_lbm/470_lbm 1.84%

Performance Improvements - Compile Time Δ Previous Current σ
SingleSource/Benchmarks/Adobe-C++/functionobjects -2.57% 0.7201 0.7016 0.0060
SingleSource/Benchmarks/Shootout-C++/fibo -2.51% 0.7045 0.6868 0.0057
SingleSource/Benchmarks/Shootout-C++/moments -2.44% 0.5503 0.5369 0.0048
SingleSource/Benchmarks/Shootout-C++/methcall -1.38% 0.7526 0.7422 0.0009
SingleSource/Benchmarks/Misc-C++/bigfib -1.31% 1.2855 1.2687 0.0032
MultiSource/Benchmarks/Prolangs-C++/employ/employ -1.31% 0.8499 0.8388 0.0039
SingleSource/Benchmarks/CoyoteBench/fftbench -1.18% 1.0390 1.0267 0.0031
MultiSource/Benchmarks/Prolangs-C++/city/city -1.00% 5.3008 5.2477 0.0100

In our internal compile time testsuite I could only see some small improvements (no regressions).

One odd thing I noticed was execution performance improvements due to this patch (multisample=5), which is probably not expected? Likely noise? but I'll paste the results just in case you want to double check those:

Execution time -O3

Performance Improvements - Execution Time Δ Previous Current σ
SingleSource/Benchmarks/Polybench/linear-algebra/kernels/trmm/trmm -4.81% 0.7553 0.7190 0.0011

Execution time -O2

Performance Improvements - Execution Time Δ Previous Current σ
SingleSource/Benchmarks/Shootout-C++/methcall -2.63% 5.1823 5.0462 0.0391
External/SPEC/CINT2000/253_perlbmk/253_perlbmk -2.14%
SingleSource/Benchmarks/Polybench/stencils/fdtd-apml/fdtd-apml -1.35% 1.1386 1.1232 0.0051

include/llvm/Analysis/MemoryDependenceAnalysis.h

 #define LLVM_ANALYSIS_MEMORYDEPENDENCEANALYSIS_H
 
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/PointerSumType.h"
 #include "llvm/ADT/PointerEmbeddedInt.h"
 #include "llvm/ADT/SmallPtrSet.h"
 #include "llvm/Analysis/AliasAnalysis.h"
 #include "llvm/IR/BasicBlock.h"
+#include "llvm/IR/PassManager.h"
 #include "llvm/IR/PredIteratorCache.h"
 #include "llvm/IR/ValueHandle.h"
 #include "llvm/Pass.h"
 
 namespace llvm {
 class Function;
 class FunctionPass;
 class Instruction;
 class CallSite;
 class AssumptionCache;
-class MemoryDependenceAnalysis;
+class MemoryDependenceResults;
 class PredIteratorCache;
 class DominatorTree;
 class PHITransAddr;
 
 /// A memory dependence query can return one of three different answers.
 class MemDepResult {
   enum DepType {
     /// Clients of MemDep never see this.
public:
   }
 
   bool operator==(const MemDepResult &M) const { return Value == M.Value; }
   bool operator!=(const MemDepResult &M) const { return Value != M.Value; }
   bool operator<(const MemDepResult &M) const { return Value < M.Value; }
   bool operator>(const MemDepResult &M) const { return Value > M.Value; }
 
 private:
-  friend class MemoryDependenceAnalysis;
+  friend class MemoryDependenceResults;
 
   /// Tests if this is a MemDepResult in its dirty/invalid. state.
   bool isDirty() const { return Value.is<Invalid>(); }
 
   static MemDepResult getDirty(Instruction *Inst) {
     return MemDepResult(ValueTy::create<Invalid>(Inst));
   }
 };
public:
   /// because of phi translation.  This returns null if the address was not
   /// available in a block (i.e. because phi translation failed) or if this is
   /// a cached result and that address was deleted.
   ///
   /// The address is always null for a non-local 'call' dependence.
   Value *getAddress() const { return Address; }
 };
 
-/// Determines, for a given memory operation, what preceding memory operations
-/// it depends on.
-///
-/// It builds on alias analysis information, and tries to provide a lazy,
-/// caching interface to a common kind of alias information query.
+/// Provides a lazy, caching interface for making common memory aliasing
+/// information queries, backed by LLVM's alias analysis passes.
 ///
 /// The dependency information returned is somewhat unusual, but is pragmatic.
 /// If queried about a store or call that might modify memory, the analysis
 /// will return the instruction[s] that may either load from that memory or
 /// store to it.  If queried with a load or call that can never modify memory,
 /// the analysis will return calls and stores that might modify the pointer,
 /// but generally does not return loads unless a) they are volatile, or
 /// b) they load from *must-aliased* pointers.  Returning a dependence on
 /// must-alias'd pointers instead of all pointers interacts well with the
 /// internal caching mechanism.
-class MemoryDependenceAnalysis : public FunctionPass {
+class MemoryDependenceResults {
   // A map from instructions to their dependency.
   typedef DenseMap<Instruction *, MemDepResult> LocalDepMapType;
   LocalDepMapType LocalDeps;
 
 public:
   typedef std::vector<NonLocalDepEntry> NonLocalDepInfo;
 
 private:
private:
   typedef DenseMap<Instruction *, SmallPtrSet<Instruction *, 4>>
       ReverseDepMapType;
   ReverseDepMapType ReverseLocalDeps;
 
   // A reverse mapping from dependencies to the non-local dependees.
   ReverseDepMapType ReverseNonLocalDeps;
 
   /// Current AA implementation, just a cache.
-  AliasAnalysis *AA;
+  AliasAnalysis &AA;
+  AssumptionCache &AC;
+  const TargetLibraryInfo &TLI;
   DominatorTree *DT;
-  AssumptionCache *AC;
-  const TargetLibraryInfo *TLI;
   PredIteratorCache PredCache;
 
 public:
-  MemoryDependenceAnalysis();
-  ~MemoryDependenceAnalysis() override;
-  static char ID;
-
-  /// Pass Implementation stuff.  This doesn't do any analysis eagerly.
-  bool runOnFunction(Function &) override;
-
-  /// Clean up memory in between runs
-  void releaseMemory() override;
-
-  /// Does not modify anything.  It uses Value Numbering and Alias Analysis.
-  void getAnalysisUsage(AnalysisUsage &AU) const override;
+  MemoryDependenceResults(AliasAnalysis &AA, AssumptionCache &AC,
+                          const TargetLibraryInfo &TLI,
+                          DominatorTree *DT = nullptr)
+      : AA(AA), AC(AC), TLI(TLI), DT(DT) {}
 
   /// Returns the instruction on which a memory operation depends.
   ///
   /// See the class comment for more details.  It is illegal to call this on
   /// non-memory instructions.
   MemDepResult getDependency(Instruction *QueryInst);
 
   /// Perform a full dependency query for the specified call, returning the set
public:
   /// that is 1) still efficient, 2) safe for the target, and 3) would provide
   /// the specified memory location value, then this function returns the size
   /// in bytes of the load width to use.  If not, this returns zero.
   static unsigned getLoadLoadClobberFullWidthSize(const Value *MemLocBase,
                                                   int64_t MemLocOffs,
                                                   unsigned MemLocSize,
                                                   const LoadInst *LI);
 
+  /// Release memory in caches.
+  void releaseMemory();
+
 private:
   MemDepResult getCallSiteDependencyFrom(CallSite C, bool isReadOnlyCall,
                                          BasicBlock::iterator ScanIt,
                                          BasicBlock *BB);
   bool getNonLocalPointerDepFromBB(Instruction *QueryInst,
                                    const PHITransAddr &Pointer,
                                    const MemoryLocation &Loc, bool isLoad,
                                    BasicBlock *BB,
                                    SmallVectorImpl<NonLocalDepResult> &Result,
                                    DenseMap<BasicBlock *, Value *> &Visited,
                                    bool SkipFirstBlock = false);
   MemDepResult GetNonLocalInfoForBlock(Instruction *QueryInst,
                                        const MemoryLocation &Loc, bool isLoad,
                                        BasicBlock *BB, NonLocalDepInfo *Cache,
                                        unsigned NumSortedEntries);
 
   void RemoveCachedNonLocalPointerDependencies(ValueIsLoadPair P);
 
   void verifyRemoved(Instruction *Inst) const;
 };
 
+/// An analysis that produces \c MemoryDependenceResults for a function.
+///
+/// This is essentially a no-op because the results are computed entirely
+/// lazily.
+struct MemoryDependenceAnalysis : AnalysisBase<MemoryDependenceAnalysis> {
+  typedef MemoryDependenceResults Result;
+
+  MemoryDependenceResults run(Function &F, AnalysisManager<Function> *AM);
+};
+
+/// A wrapper analysis pass for the legacy pass manager that exposes a \c
+/// MemoryDepnedenceResults instance.
+class MemoryDependenceWrapperPass : public FunctionPass {
+  Optional<MemoryDependenceResults> MemDep;
+public:
+  MemoryDependenceWrapperPass();
+  ~MemoryDependenceWrapperPass() override;
+  static char ID;
+
+  /// Pass Implementation stuff.  This doesn't do any analysis eagerly.
+  bool runOnFunction(Function &) override;
+
+  /// Clean up memory in between runs
+  void releaseMemory() override;
+
+  /// Does not modify anything.  It uses Value Numbering and Alias Analysis.
+  void getAnalysisUsage(AnalysisUsage &AU) const override;
+
+  MemoryDependenceResults &getMemDep() { return *MemDep; }
+};
+
 } // End llvm namespace
 
 #endif

include/llvm/InitializePasses.h

 void initializeMachineRegionInfoPassPass(PassRegistry&);
 void initializeMachineSchedulerPass(PassRegistry&);
 void initializeMachineSinkingPass(PassRegistry&);
 void initializeMachineTraceMetricsPass(PassRegistry&);
 void initializeMachineVerifierPassPass(PassRegistry&);
 void initializeMemCpyOptPass(PassRegistry&);
 void initializeMemDepPrinterPass(PassRegistry&);
 void initializeMemDerefPrinterPass(PassRegistry&);
-void initializeMemoryDependenceAnalysisPass(PassRegistry&);
+void initializeMemoryDependenceWrapperPassPass(PassRegistry&);
 void initializeMemorySSALazyPass(PassRegistry&);
 void initializeMemorySSAPrinterPassPass(PassRegistry&);
 void initializeMergedLoadStoreMotionPass(PassRegistry &);
 void initializeMetaRenamerPass(PassRegistry&);
 void initializeMergeFunctionsPass(PassRegistry&);
 void initializeModuleDebugInfoPrinterPass(PassRegistry&);
 void initializeNaryReassociatePass(PassRegistry&);
 void initializeNoAAPass(PassRegistry&);

include/llvm/Transforms/Utils/BasicBlockUtils.h

 
 // FIXME: Move to this file: BasicBlock::removePredecessor, BB::splitBasicBlock
 
 #include "llvm/IR/BasicBlock.h"
 #include "llvm/IR/CFG.h"
 
 namespace llvm {
 
-class MemoryDependenceAnalysis;
+class MemoryDependenceResults;
 class DominatorTree;
 class LoopInfo;
 class Instruction;
 class MDNode;
 class ReturnInst;
 class TargetLibraryInfo;
 class TerminatorInst;
 
 /// DeleteDeadBlock - Delete the specified block, which must have no
 /// predecessors.
 void DeleteDeadBlock(BasicBlock *BB);
 
 /// FoldSingleEntryPHINodes - We know that BB has one predecessor.  If there are
 /// any single-entry PHI nodes in it, fold them away.  This handles the case
 /// when all entries to the PHI nodes in a block are guaranteed equal, such as
 /// when the block has exactly one predecessor.
 void FoldSingleEntryPHINodes(BasicBlock *BB,
-                             MemoryDependenceAnalysis *MemDep = nullptr);
+                             MemoryDependenceResults *MemDep = nullptr);
 
 /// DeleteDeadPHIs - Examine each PHI in the given block and delete it if it
 /// is dead. Also recursively delete any operands that become dead as
 /// a result. This includes tracing the def-use list from the PHI to see if
 /// it is ultimately unused or if it reaches an unused cycle. Return true
 /// if any PHIs were deleted.
 bool DeleteDeadPHIs(BasicBlock *BB, const TargetLibraryInfo *TLI = nullptr);
 
 /// MergeBlockIntoPredecessor - Attempts to merge a block into its predecessor,
 /// if possible.  The return value indicates success or failure.
 bool MergeBlockIntoPredecessor(BasicBlock *BB, DominatorTree *DT = nullptr,
                                LoopInfo *LI = nullptr,
-                               MemoryDependenceAnalysis *MemDep = nullptr);
+                               MemoryDependenceResults *MemDep = nullptr);
 
 // ReplaceInstWithValue - Replace all uses of an instruction (specified by BI)
 // with a value, then remove and delete the original instruction.
 //
 void ReplaceInstWithValue(BasicBlock::InstListType &BIL,
                           BasicBlock::iterator &BI, Value *V);
 
 // ReplaceInstWithInst - Replace the instruction specified by BI with the

lib/Analysis/Analysis.cpp

void llvm::initializeAnalysis(PassRegistry &Registry) {
   initializeIVUsersPass(Registry);
   initializeInstCountPass(Registry);
   initializeIntervalPartitionPass(Registry);
   initializeLazyValueInfoPass(Registry);
   initializeLintPass(Registry);
   initializeLoopInfoWrapperPassPass(Registry);
   initializeMemDepPrinterPass(Registry);
   initializeMemDerefPrinterPass(Registry);
-  initializeMemoryDependenceAnalysisPass(Registry);
+  initializeMemoryDependenceWrapperPassPass(Registry);
   initializeModuleDebugInfoPrinterPass(Registry);
   initializeObjCARCAAWrapperPassPass(Registry);
   initializePostDominatorTreeWrapperPassPass(Registry);
   initializeRegionInfoPassPass(Registry);
   initializeRegionViewerPass(Registry);
   initializeRegionPrinterPass(Registry);
   initializeRegionOnlyViewerPass(Registry);
   initializeRegionOnlyPrinterPass(Registry);

lib/Analysis/MemDepPrinter.cpp

struct MemDepPrinter : public FunctionPass {
     }
 
     bool runOnFunction(Function &F) override;
 
     void print(raw_ostream &OS, const Module * = nullptr) const override;
 
     void getAnalysisUsage(AnalysisUsage &AU) const override {
       AU.addRequiredTransitive<AAResultsWrapperPass>();
-      AU.addRequiredTransitive<MemoryDependenceAnalysis>();
+      AU.addRequiredTransitive<MemoryDependenceWrapperPass>();
       AU.setPreservesAll();
     }
 
     void releaseMemory() override {
       Deps.clear();
       F = nullptr;
     }
 
static InstTypePair getInstTypePair(const Instruction* inst, DepType type) {
       return InstTypePair(inst, type);
     }
   };
 }
 
 char MemDepPrinter::ID = 0;
 INITIALIZE_PASS_BEGIN(MemDepPrinter, "print-memdeps",
                       "Print MemDeps of function", false, true)
-INITIALIZE_PASS_DEPENDENCY(MemoryDependenceAnalysis)
+INITIALIZE_PASS_DEPENDENCY(MemoryDependenceWrapperPass)
 INITIALIZE_PASS_END(MemDepPrinter, "print-memdeps",
                       "Print MemDeps of function", false, true)
 
 FunctionPass *llvm::createMemDepPrinter() {
   return new MemDepPrinter();
 }
 
 const char *const MemDepPrinter::DepTypeStr[]
   = {"Clobber", "Def", "NonFuncLocal", "Unknown"};
 
 bool MemDepPrinter::runOnFunction(Function &F) {
   this->F = &F;
-  MemoryDependenceAnalysis &MDA = getAnalysis<MemoryDependenceAnalysis>();
+  MemoryDependenceResults &MDA = getAnalysis<MemoryDependenceWrapperPass>().getMemDep();
 
   // All this code uses non-const interfaces because MemDep is not
   // const-friendly, though nothing is actually modified.
   for (auto &I : instructions(F)) {
     Instruction *Inst = &I;
 
     if (!Inst->mayReadFromMemory() && !Inst->mayWriteToMemory())
       continue;
 
     MemDepResult Res = MDA.getDependency(Inst);
     if (!Res.isNonLocal()) {
       Deps[Inst].insert(std::make_pair(getInstTypePair(Res),
                                        static_cast<BasicBlock *>(nullptr)));
     } else if (auto CS = CallSite(Inst)) {
-      const MemoryDependenceAnalysis::NonLocalDepInfo &NLDI =
+      const MemoryDependenceResults::NonLocalDepInfo &NLDI =
         MDA.getNonLocalCallDependency(CS);
 
       DepSet &InstDeps = Deps[Inst];
-      for (MemoryDependenceAnalysis::NonLocalDepInfo::const_iterator
+      for (MemoryDependenceResults::NonLocalDepInfo::const_iterator
            I = NLDI.begin(), E = NLDI.end(); I != E; ++I) {
         const MemDepResult &Res = I->getResult();
         InstDeps.insert(std::make_pair(getInstTypePair(Res), I->getBB()));
       }
     } else {
       SmallVector<NonLocalDepResult, 4> NLDI;
       assert( (isa<LoadInst>(Inst) || isa<StoreInst>(Inst) ||
                isa<VAArgInst>(Inst)) && "Unknown memory instruction!"); 

lib/Analysis/MemoryDependenceAnalysis.cpp

 static cl::opt<unsigned>
     BlockNumberLimit("memdep-block-number-limit", cl::Hidden, cl::init(1000),
                      cl::desc("The number of blocks to scan during memory "
                               "dependency analysis (default = 1000)"));
 
 // Limit on the number of memdep results to process.
 static const unsigned int NumResultsLimit = 100;
 
-char MemoryDependenceAnalysis::ID = 0;
-
-INITIALIZE_PASS_BEGIN(MemoryDependenceAnalysis, "memdep",
-                      "Memory Dependence Analysis", false, true)
-INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker)
-INITIALIZE_PASS_DEPENDENCY(AAResultsWrapperPass)
-INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)
-INITIALIZE_PASS_END(MemoryDependenceAnalysis, "memdep",
-                    "Memory Dependence Analysis", false, true)
-
-MemoryDependenceAnalysis::MemoryDependenceAnalysis() : FunctionPass(ID) {
-  initializeMemoryDependenceAnalysisPass(*PassRegistry::getPassRegistry());
-}
-MemoryDependenceAnalysis::~MemoryDependenceAnalysis() {}
-
-/// Clean up memory in between runs
-void MemoryDependenceAnalysis::releaseMemory() {
-  LocalDeps.clear();
-  NonLocalDeps.clear();
-  NonLocalPointerDeps.clear();
-  ReverseLocalDeps.clear();
-  ReverseNonLocalDeps.clear();
-  ReverseNonLocalPtrDeps.clear();
-  PredCache.clear();
-}
-
-void MemoryDependenceAnalysis::getAnalysisUsage(AnalysisUsage &AU) const {
-  AU.setPreservesAll();
-  AU.addRequired<AssumptionCacheTracker>();
-  AU.addRequiredTransitive<AAResultsWrapperPass>();
-  AU.addRequiredTransitive<TargetLibraryInfoWrapperPass>();
-}
-
-bool MemoryDependenceAnalysis::runOnFunction(Function &F) {
-  AA = &getAnalysis<AAResultsWrapperPass>().getAAResults();
-  AC = &getAnalysis<AssumptionCacheTracker>().getAssumptionCache(F);
-  DominatorTreeWrapperPass *DTWP =
-      getAnalysisIfAvailable<DominatorTreeWrapperPass>();
-  DT = DTWP ? &DTWP->getDomTree() : nullptr;
-  TLI = &getAnalysis<TargetLibraryInfoWrapperPass>().getTLI();
-  return false;
-}
-
 /// This is a helper function that removes Val from 'Inst's set in ReverseMap.
 ///
 /// If the set becomes empty, remove Inst's entry.
 template <typename KeyTy>
 static void
 RemoveFromReverseMap(DenseMap<Instruction *, SmallPtrSet<KeyTy, 4>> &ReverseMap,
                      Instruction *Inst, KeyTy Val) {
   typename DenseMap<Instruction *, SmallPtrSet<KeyTy, 4>>::iterator InstIt =
static ModRefInfo GetLocation(const Instruction *Inst, MemoryLocation &Loc,
   if (Inst->mayWriteToMemory())
     return MRI_ModRef;
   if (Inst->mayReadFromMemory())
     return MRI_Ref;
   return MRI_NoModRef;
 }
 
 /// Private helper for finding the local dependencies of a call site.
-MemDepResult MemoryDependenceAnalysis::getCallSiteDependencyFrom(
+MemDepResult MemoryDependenceResults::getCallSiteDependencyFrom(
     CallSite CS, bool isReadOnlyCall, BasicBlock::iterator ScanIt,
     BasicBlock *BB) {
   unsigned Limit = BlockScanLimit;
 
   // Walk backwards through the block, looking for dependencies
   while (ScanIt != BB->begin()) {
     // Limit the amount of scanning we do so we don't end up with quadratic
     // running time on extreme testcases.
     --Limit;
     if (!Limit)
       return MemDepResult::getUnknown();
 
     Instruction *Inst = &*--ScanIt;
 
     // If this inst is a memory op, get the pointer it accessed
     MemoryLocation Loc;
-    ModRefInfo MR = GetLocation(Inst, Loc, *TLI);
+    ModRefInfo MR = GetLocation(Inst, Loc, TLI);
     if (Loc.Ptr) {
       // A simple instruction.
-      if (AA->getModRefInfo(CS, Loc) != MRI_NoModRef)
+      if (AA.getModRefInfo(CS, Loc) != MRI_NoModRef)
         return MemDepResult::getClobber(Inst);
       continue;
     }
 
     if (auto InstCS = CallSite(Inst)) {
       // Debug intrinsics don't cause dependences.
       if (isa<DbgInfoIntrinsic>(Inst))
         continue;
       // If these two calls do not interfere, look past it.
-      switch (AA->getModRefInfo(CS, InstCS)) {
+      switch (AA.getModRefInfo(CS, InstCS)) {
       case MRI_NoModRef:
         // If the two calls are the same, return InstCS as a Def, so that
         // CS can be found redundant and eliminated.
         if (isReadOnlyCall && !(MR & MRI_Mod) &&
             CS.getInstruction()->isIdenticalToWhenDefined(Inst))
           return MemDepResult::getDef(Inst);
 
         // Otherwise if the two calls don't interact (e.g. InstCS is readnone)
static bool isLoadLoadClobberIfExtendedToFullWidth(const MemoryLocation &MemLoc,
                                                    int64_t &MemLocOffs,
                                                    const LoadInst *LI) {
   const DataLayout &DL = LI->getModule()->getDataLayout();
 
   // If we haven't already computed the base/offset of MemLoc, do so now.
   if (!MemLocBase)
     MemLocBase = GetPointerBaseWithConstantOffset(MemLoc.Ptr, MemLocOffs, DL);
 
-  unsigned Size = MemoryDependenceAnalysis::getLoadLoadClobberFullWidthSize(
+  unsigned Size = MemoryDependenceResults::getLoadLoadClobberFullWidthSize(
       MemLocBase, MemLocOffs, MemLoc.Size, LI);
   return Size != 0;
 }
 
-unsigned MemoryDependenceAnalysis::getLoadLoadClobberFullWidthSize(
+unsigned MemoryDependenceResults::getLoadLoadClobberFullWidthSize(
     const Value *MemLocBase, int64_t MemLocOffs, unsigned MemLocSize,
     const LoadInst *LI) {
   // We can only extend simple integer loads.
   if (!isa<IntegerType>(LI->getType()) || !LI->isSimple())
     return 0;
 
   // Load widening is hostile to ThreadSanitizer: it may cause false positives
   // or make the reports more cryptic (access sizes are wrong).
if (LoadInst *LI = dyn_cast<LoadInst>(Inst))
     return LI->isVolatile();
   else if (StoreInst *SI = dyn_cast<StoreInst>(Inst))
     return SI->isVolatile();
   else if (AtomicCmpXchgInst *AI = dyn_cast<AtomicCmpXchgInst>(Inst))
     return AI->isVolatile();
   return false;
 }
 
-MemDepResult MemoryDependenceAnalysis::getPointerDependencyFrom(
+MemDepResult MemoryDependenceResults::getPointerDependencyFrom(
     const MemoryLocation &MemLoc, bool isLoad, BasicBlock::iterator ScanIt,
     BasicBlock *BB, Instruction *QueryInst) {
 
   if (QueryInst != nullptr) {
     if (auto *LI = dyn_cast<LoadInst>(QueryInst)) {
       MemDepResult invariantGroupDependency =
           getInvariantGroupPointerDependency(LI, BB);
 
       if (invariantGroupDependency.isDef())
         return invariantGroupDependency;
     }
   }
   return getSimplePointerDependencyFrom(MemLoc, isLoad, ScanIt, BB, QueryInst);
 }
 
 MemDepResult
-MemoryDependenceAnalysis::getInvariantGroupPointerDependency(LoadInst *LI,
+MemoryDependenceResults::getInvariantGroupPointerDependency(LoadInst *LI,
                                                              BasicBlock *BB) {
   Value *LoadOperand = LI->getPointerOperand();
   // It's is not safe to walk the use list of global value, because function
   // passes aren't allowed to look outside their functions.
   if (isa<GlobalValue>(LoadOperand))
     return MemDepResult::getUnknown();
 
   auto *InvariantGroupMD = LI->getMetadata(LLVMContext::MD_invariant_group);
for (Use &Us : Ptr->uses()) {
       if ((isa<LoadInst>(U) || isa<StoreInst>(U)) && U->getParent() == BB &&
           U->getMetadata(LLVMContext::MD_invariant_group) == InvariantGroupMD)
         return MemDepResult::getDef(U);
     }
   }
   return Result;
 }
 
-MemDepResult MemoryDependenceAnalysis::getSimplePointerDependencyFrom(
+MemDepResult MemoryDependenceResults::getSimplePointerDependencyFrom(
     const MemoryLocation &MemLoc, bool isLoad, BasicBlock::iterator ScanIt,
     BasicBlock *BB, Instruction *QueryInst) {
 
   const Value *MemLocBase = nullptr;
   int64_t MemLocOffset = 0;
   unsigned Limit = BlockScanLimit;
   bool isInvariantLoad = false;
 
while (ScanIt != BB->begin()) {
     if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(Inst)) {
       // If we reach a lifetime begin or end marker, then the query ends here
       // because the value is undefined.
       if (II->getIntrinsicID() == Intrinsic::lifetime_start) {
         // FIXME: This only considers queries directly on the invariant-tagged
         // pointer, not on query pointers that are indexed off of them.  It'd
         // be nice to handle that at some point (the right approach is to use
         // GetPointerBaseWithConstantOffset).
-        if (AA->isMustAlias(MemoryLocation(II->getArgOperand(1)), MemLoc))
+        if (AA.isMustAlias(MemoryLocation(II->getArgOperand(1)), MemLoc))
           return MemDepResult::getDef(II);
         continue;
       }
     }
 
     // Values depend on loads if the pointers are must aliased.  This means
     // that a load depends on another must aliased load from the same value.
     // One exception is atomic loads: a value can depend on an atomic load that
if (LoadInst *LI = dyn_cast<LoadInst>(Inst)) {
           return MemDepResult::getClobber(LI);
         if (LI->getOrdering() != Monotonic)
           return MemDepResult::getClobber(LI);
       }
 
       MemoryLocation LoadLoc = MemoryLocation::get(LI);
 
       // If we found a pointer, check if it could be the same as our pointer.
-      AliasResult R = AA->alias(LoadLoc, MemLoc);
+      AliasResult R = AA.alias(LoadLoc, MemLoc);
 
       if (isLoad) {
         if (R == NoAlias) {
           // If this is an over-aligned integer load (for example,
           // "load i8* %P, align 4") see if it would obviously overlap with the
           // queried location if widened to a larger load (e.g. if the queried
           // location is 1 byte at P+1).  If so, return it as a load/load
           // clobber result, allowing the client to decide to widen the load if
#endif
         continue;
       }
 
       // Stores don't depend on other no-aliased accesses.
       if (R == NoAlias)
         continue;
 
       // Stores don't alias loads from read-only memory.
-      if (AA->pointsToConstantMemory(LoadLoc))
+      if (AA.pointsToConstantMemory(LoadLoc))
         continue;
 
       // Stores depend on may/must aliased loads.
       return MemDepResult::getDef(Inst);
     }
 
     if (StoreInst *SI = dyn_cast<StoreInst>(Inst)) {
       // Atomic stores have complications involved.
if (StoreInst *SI = dyn_cast<StoreInst>(Inst)) {
       if (SI->isVolatile())
         if (!QueryInst || isNonSimpleLoadOrStore(QueryInst) ||
             isOtherMemAccess(QueryInst))
           return MemDepResult::getClobber(SI);
 
       // If alias analysis can tell that this store is guaranteed to not modify
       // the query pointer, ignore it.  Use getModRefInfo to handle cases where
       // the query pointer points to constant memory etc.
-      if (AA->getModRefInfo(SI, MemLoc) == MRI_NoModRef)
+      if (AA.getModRefInfo(SI, MemLoc) == MRI_NoModRef)
         continue;
 
       // Ok, this store might clobber the query pointer.  Check to see if it is
       // a must alias: in this case, we want to return this as a def.
       MemoryLocation StoreLoc = MemoryLocation::get(SI);
 
       // If we found a pointer, check if it could be the same as our pointer.
-      AliasResult R = AA->alias(StoreLoc, MemLoc);
+      AliasResult R = AA.alias(StoreLoc, MemLoc);
 
       if (R == NoAlias)
         continue;
       if (R == MustAlias)
         return MemDepResult::getDef(Inst);
       if (isInvariantLoad)
         continue;
       return MemDepResult::getClobber(Inst);
     }
 
     // If this is an allocation, and if we know that the accessed pointer is to
     // the allocation, return Def.  This means that there is no dependence and
     // the access can be optimized based on that.  For example, a load could
     // turn into undef.
     // Note: Only determine this to be a malloc if Inst is the malloc call, not
     // a subsequent bitcast of the malloc call result.  There can be stores to
     // the malloced memory between the malloc call and its bitcast uses, and we
     // need to continue scanning until the malloc call.
-    if (isa<AllocaInst>(Inst) || isNoAliasFn(Inst, TLI)) {
+    if (isa<AllocaInst>(Inst) || isNoAliasFn(Inst, &TLI)) {
       const Value *AccessPtr = GetUnderlyingObject(MemLoc.Ptr, DL);
 
-      if (AccessPtr == Inst || AA->isMustAlias(Inst, AccessPtr))
+      if (AccessPtr == Inst || AA.isMustAlias(Inst, AccessPtr))
         return MemDepResult::getDef(Inst);
       if (isInvariantLoad)
         continue;
       // Be conservative if the accessed pointer may alias the allocation -
       // fallback to the generic handling below.
-      if ((AA->alias(Inst, AccessPtr) == NoAlias) &&
+      if ((AA.alias(Inst, AccessPtr) == NoAlias) &&
           // If the allocation is not aliased and does not read memory (like
           // strdup), it is safe to ignore.
-          (isa<AllocaInst>(Inst) || isMallocLikeFn(Inst, TLI) ||
-           isCallocLikeFn(Inst, TLI)))
+          (isa<AllocaInst>(Inst) || isMallocLikeFn(Inst, &TLI) ||
+           isCallocLikeFn(Inst, &TLI)))
         continue;
     }
 
     if (isInvariantLoad)
       continue;
 
     // See if this instruction (e.g. a call or vaarg) mod/ref's the pointer.
-    ModRefInfo MR = AA->getModRefInfo(Inst, MemLoc);
+    ModRefInfo MR = AA.getModRefInfo(Inst, MemLoc);
     // If necessary, perform additional analysis.
     if (MR == MRI_ModRef)
-      MR = AA->callCapturesBefore(Inst, MemLoc, DT, &OBB);
+      MR = AA.callCapturesBefore(Inst, MemLoc, DT, &OBB);
     switch (MR) {
     case MRI_NoModRef:
       // If the call has no effect on the queried pointer, just ignore it.
       continue;
     case MRI_Mod:
       return MemDepResult::getClobber(Inst);
     case MRI_Ref:
       // If the call is known to never store to the pointer, and if this is a
       // load query, we can safely ignore it (scan past it).
       if (isLoad)
         continue;
     default:
       // Otherwise, there is a potential dependence.  Return a clobber.
       return MemDepResult::getClobber(Inst);
     }
   }
 
   // No dependence found.  If this is the entry block of the function, it is
   // unknown, otherwise it is non-local.
   if (BB != &BB->getParent()->getEntryBlock())
     return MemDepResult::getNonLocal();
   return MemDepResult::getNonFuncLocal();
 }
 
-MemDepResult MemoryDependenceAnalysis::getDependency(Instruction *QueryInst) {
+MemDepResult MemoryDependenceResults::getDependency(Instruction *QueryInst) {
   Instruction *ScanPos = QueryInst;
 
   // Check for a cached result
   MemDepResult &LocalCache = LocalDeps[QueryInst];
 
   // If the cached entry is non-dirty, just return it.  Note that this depends
   // on MemDepResult's default constructing to 'dirty'.
   if (!LocalCache.isDirty())
if (BasicBlock::iterator(QueryInst) == QueryParent->begin()) {
     // No dependence found.  If this is the entry block of the function, it is
     // unknown, otherwise it is non-local.
     if (QueryParent != &QueryParent->getParent()->getEntryBlock())
       LocalCache = MemDepResult::getNonLocal();
     else
       LocalCache = MemDepResult::getNonFuncLocal();
   } else {
     MemoryLocation MemLoc;
-    ModRefInfo MR = GetLocation(QueryInst, MemLoc, *TLI);
+    ModRefInfo MR = GetLocation(QueryInst, MemLoc, TLI);
     if (MemLoc.Ptr) {
       // If we can do a pointer scan, make it happen.
       bool isLoad = !(MR & MRI_Mod);
       if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(QueryInst))
         isLoad |= II->getIntrinsicID() == Intrinsic::lifetime_start;
 
       LocalCache = getPointerDependencyFrom(
           MemLoc, isLoad, ScanPos->getIterator(), QueryParent, QueryInst);
     } else if (isa<CallInst>(QueryInst) || isa<InvokeInst>(QueryInst)) {
       CallSite QueryCS(QueryInst);
-      bool isReadOnly = AA->onlyReadsMemory(QueryCS);
+      bool isReadOnly = AA.onlyReadsMemory(QueryCS);
       LocalCache = getCallSiteDependencyFrom(
           QueryCS, isReadOnly, ScanPos->getIterator(), QueryParent);
     } else
       // Non-memory instruction.
       LocalCache = MemDepResult::getUnknown();
   }
 
   // Remember the result!
   if (Instruction *I = LocalCache.getInst())
     ReverseLocalDeps[I].insert(QueryInst);
 
   return LocalCache;
 }
 
 #ifndef NDEBUG
 /// This method is used when -debug is specified to verify that cache arrays
 /// are properly kept sorted.
-static void AssertSorted(MemoryDependenceAnalysis::NonLocalDepInfo &Cache,
+static void AssertSorted(MemoryDependenceResults::NonLocalDepInfo &Cache,
                          int Count = -1) {
   if (Count == -1)
     Count = Cache.size();
   assert(std::is_sorted(Cache.begin(), Cache.begin() + Count) &&
          "Cache isn't sorted!");
 }
 #endif
 
-const MemoryDependenceAnalysis::NonLocalDepInfo &
-MemoryDependenceAnalysis::getNonLocalCallDependency(CallSite QueryCS) {
+const MemoryDependenceResults::NonLocalDepInfo &
+MemoryDependenceResults::getNonLocalCallDependency(CallSite QueryCS) {
   assert(getDependency(QueryCS.getInstruction()).isNonLocal() &&
          "getNonLocalCallDependency should only be used on calls with "
          "non-local deps!");
   PerInstNLInfo &CacheP = NonLocalDeps[QueryCS.getInstruction()];
   NonLocalDepInfo &Cache = CacheP.first;
 
   // This is the set of blocks that need to be recomputed.  In the cached case,
   // this can happen due to instructions being deleted etc. In the uncached
if (!Cache.empty()) {
     // Seed DirtyBlocks with each of the preds of QueryInst's block.
     BasicBlock *QueryBB = QueryCS.getInstruction()->getParent();
     for (BasicBlock *Pred : PredCache.get(QueryBB))
       DirtyBlocks.push_back(Pred);
     ++NumUncacheNonLocal;
   }
 
   // isReadonlyCall - If this is a read-only call, we can be more aggressive.
-  bool isReadonlyCall = AA->onlyReadsMemory(QueryCS);
+  bool isReadonlyCall = AA.onlyReadsMemory(QueryCS);
 
   SmallPtrSet<BasicBlock *, 32> Visited;
 
   unsigned NumSortedEntries = Cache.size();
   DEBUG(AssertSorted(Cache));
 
   // Iterate while we still have blocks to update.
   while (!DirtyBlocks.empty()) {
if (!Dep.isNonLocal()) {
       for (BasicBlock *Pred : PredCache.get(DirtyBB))
         DirtyBlocks.push_back(Pred);
     }
   }
 
   return Cache;
 }
 
-void MemoryDependenceAnalysis::getNonLocalPointerDependency(
+void MemoryDependenceResults::getNonLocalPointerDependency(
     Instruction *QueryInst, SmallVectorImpl<NonLocalDepResult> &Result) {
   const MemoryLocation Loc = MemoryLocation::get(QueryInst);
   bool isLoad = isa<LoadInst>(QueryInst);
   BasicBlock *FromBB = QueryInst->getParent();
   assert(FromBB);
 
   assert(Loc.Ptr->getType()->isPointerTy() &&
          "Can't get pointer deps of a non-pointer!");
auto isOrdered = [](Instruction *Inst) {
     return false;
   };
   if (isVolatile(QueryInst) || isOrdered(QueryInst)) {
     Result.push_back(NonLocalDepResult(FromBB, MemDepResult::getUnknown(),
                                        const_cast<Value *>(Loc.Ptr)));
     return;
   }
   const DataLayout &DL = FromBB->getModule()->getDataLayout();
-  PHITransAddr Address(const_cast<Value *>(Loc.Ptr), DL, AC);
+  PHITransAddr Address(const_cast<Value *>(Loc.Ptr), DL, &AC);
 
   // This is the set of blocks we've inspected, and the pointer we consider in
   // each block.  Because of critical edges, we currently bail out if querying
   // a block with multiple different pointers.  This can happen during PHI
   // translation.
   DenseMap<BasicBlock *, Value *> Visited;
   if (getNonLocalPointerDepFromBB(QueryInst, Address, Loc, isLoad, FromBB,
                                    Result, Visited, true))
     return;
   Result.clear();
   Result.push_back(NonLocalDepResult(FromBB, MemDepResult::getUnknown(),
                                      const_cast<Value *>(Loc.Ptr)));
 }
 
 /// Compute the memdep value for BB with Pointer/PointeeSize using either
 /// cached information in Cache or by doing a lookup (which may use dirty cache
 /// info if available).
 ///
 /// If we do a lookup, add the result to the cache.
-MemDepResult MemoryDependenceAnalysis::GetNonLocalInfoForBlock(
+MemDepResult MemoryDependenceResults::GetNonLocalInfoForBlock(
     Instruction *QueryInst, const MemoryLocation &Loc, bool isLoad,
     BasicBlock *BB, NonLocalDepInfo *Cache, unsigned NumSortedEntries) {
 
   // Do a binary search to see if we already have an entry for this block in
   // the cache set.  If so, find it.
   NonLocalDepInfo::iterator Entry = std::upper_bound(
       Cache->begin(), Cache->begin() + NumSortedEntries, NonLocalDepEntry(BB));
   if (Entry != Cache->begin() && (Entry - 1)->getBB() == BB)
MemDepResult MemoryDependenceResults::GetNonLocalInfoForBlock(
   return Dep;
 }
 
 /// Sort the NonLocalDepInfo cache, given a certain number of elements in the
 /// array that are already properly ordered.
 ///
 /// This is optimized for the case when only a few entries are added.
 static void
-SortNonLocalDepInfoCache(MemoryDependenceAnalysis::NonLocalDepInfo &Cache,
+SortNonLocalDepInfoCache(MemoryDependenceResults::NonLocalDepInfo &Cache,
                          unsigned NumSortedEntries) {
   switch (Cache.size() - NumSortedEntries) {
   case 0:
     // done, no new entries.
     break;
   case 2: {
     // Two new entries, insert the last one into place.
     NonLocalDepEntry Val = Cache.back();
     Cache.pop_back();
-    MemoryDependenceAnalysis::NonLocalDepInfo::iterator Entry =
+    MemoryDependenceResults::NonLocalDepInfo::iterator Entry =
         std::upper_bound(Cache.begin(), Cache.end() - 1, Val);
     Cache.insert(Entry, Val);
     // FALL THROUGH.
   }
   case 1:
     // One new entry, Just insert the new value at the appropriate position.
     if (Cache.size() != 1) {
       NonLocalDepEntry Val = Cache.back();
       Cache.pop_back();
-      MemoryDependenceAnalysis::NonLocalDepInfo::iterator Entry =
+      MemoryDependenceResults::NonLocalDepInfo::iterator Entry =
           std::upper_bound(Cache.begin(), Cache.end(), Val);
       Cache.insert(Entry, Val);
     }
     break;
   default:
     // Added many values, do a full scale sort.
     std::sort(Cache.begin(), Cache.end());
     break;
   }
 }
 
 /// Perform a dependency query based on pointer/pointeesize starting at the end
 /// of StartBB.
 ///
 /// Add any clobber/def results to the results vector and keep track of which
 /// blocks are visited in 'Visited'.
 ///
 /// This has special behavior for the first block queries (when SkipFirstBlock
 /// is true).  In this special case, it ignores the contents of the specified
 /// block and starts returning dependence info for its predecessors.
 ///
 /// This function returns true on success, or false to indicate that it could
 /// not compute dependence information for some reason.  This should be treated
 /// as a clobber dependence on the first instruction in the predecessor block.
-bool MemoryDependenceAnalysis::getNonLocalPointerDepFromBB(
+bool MemoryDependenceResults::getNonLocalPointerDepFromBB(
     Instruction *QueryInst, const PHITransAddr &Pointer,
     const MemoryLocation &Loc, bool isLoad, BasicBlock *StartBB,
     SmallVectorImpl<NonLocalDepResult> &Result,
     DenseMap<BasicBlock *, Value *> &Visited, bool SkipFirstBlock) {
   // Look up the cached info for Pointer.
   ValueIsLoadPair CacheKey(Pointer.getAddr(), isLoad);
 
   // Set up a temporary NLPI value. If the map doesn't yet have an entry for
bool MemoryDependenceResults::getNonLocalPointerDepFromBB(
 
   // Okay, we're done now.  If we added new values to the cache, re-sort it.
   SortNonLocalDepInfoCache(*Cache, NumSortedEntries);
   DEBUG(AssertSorted(*Cache));
   return true;
 }
 
 /// If P exists in CachedNonLocalPointerInfo, remove it.
-void MemoryDependenceAnalysis::RemoveCachedNonLocalPointerDependencies(
+void MemoryDependenceResults::RemoveCachedNonLocalPointerDependencies(
     ValueIsLoadPair P) {
   CachedNonLocalPointerInfo::iterator It = NonLocalPointerDeps.find(P);
   if (It == NonLocalPointerDeps.end())
     return;
 
   // Remove all of the entries in the BB->val map.  This involves removing
   // instructions from the reverse map.
   NonLocalDepInfo &PInfo = It->second.NonLocalDeps;
 
   for (unsigned i = 0, e = PInfo.size(); i != e; ++i) {
     Instruction *Target = PInfo[i].getResult().getInst();
     if (!Target)
       continue; // Ignore non-local dep results.
     assert(Target->getParent() == PInfo[i].getBB());
 
     // Eliminating the dirty entry from 'Cache', so update the reverse info.
     RemoveFromReverseMap(ReverseNonLocalPtrDeps, Target, P);
   }
 
   // Remove P from NonLocalPointerDeps (which deletes NonLocalDepInfo).
   NonLocalPointerDeps.erase(It);
 }
 
-void MemoryDependenceAnalysis::invalidateCachedPointerInfo(Value *Ptr) {
+void MemoryDependenceResults::invalidateCachedPointerInfo(Value *Ptr) {
   // If Ptr isn't really a pointer, just ignore it.
   if (!Ptr->getType()->isPointerTy())
     return;
   // Flush store info for the pointer.
   RemoveCachedNonLocalPointerDependencies(ValueIsLoadPair(Ptr, false));
   // Flush load info for the pointer.
   RemoveCachedNonLocalPointerDependencies(ValueIsLoadPair(Ptr, true));
 }
 
-void MemoryDependenceAnalysis::invalidateCachedPredecessors() {
+void MemoryDependenceResults::invalidateCachedPredecessors() {
   PredCache.clear();
 }
 
-void MemoryDependenceAnalysis::removeInstruction(Instruction *RemInst) {
+void MemoryDependenceResults::removeInstruction(Instruction *RemInst) {
   // Walk through the Non-local dependencies, removing this one as the value
   // for any cached queries.
   NonLocalDepMapType::iterator NLDI = NonLocalDeps.find(RemInst);
   if (NLDI != NonLocalDeps.end()) {
     NonLocalDepInfo &BlockMap = NLDI->second.first;
     for (auto &Entry : BlockMap)
       if (Instruction *Inst = Entry.getResult().getInst())
         RemoveFromReverseMap(ReverseNonLocalDeps, Inst, RemInst);
void MemoryDependenceResults::removeInstruction(Instruction *RemInst) {
   assert(!NonLocalDeps.count(RemInst) && "RemInst got reinserted?");
   DEBUG(verifyRemoved(RemInst));
 }
 
 /// Verify that the specified instruction does not occur in our internal data
 /// structures.
 ///
 /// This function verifies by asserting in debug builds.
-void MemoryDependenceAnalysis::verifyRemoved(Instruction *D) const {
+void MemoryDependenceResults::verifyRemoved(Instruction *D) const {
 #ifndef NDEBUG
   for (const auto &DepKV : LocalDeps) {
     assert(DepKV.first != D && "Inst occurs in data structures");
     assert(DepKV.second.getInst() != D && "Inst occurs in data structures");
   }
 
   for (const auto &DepKV : NonLocalPointerDeps) {
     assert(DepKV.first.getPointer() != D && "Inst occurs in NLPD map key");
for (const auto &DepKV : ReverseNonLocalPtrDeps) {
     assert(DepKV.first != D && "Inst occurs in rev NLPD map");
 
     for (ValueIsLoadPair P : DepKV.second)
       assert(P != ValueIsLoadPair(D, false) && P != ValueIsLoadPair(D, true) &&
              "Inst occurs in ReverseNonLocalPtrDeps map");
   }
 #endif
 }
+
+MemoryDependenceResults
+MemoryDependenceAnalysis::run(Function &F, AnalysisManager<Function> *AM) {
+  auto &AA = AM->getResult<AAManager>(F);
+  auto &AC = AM->getResult<AssumptionAnalysis>(F);
+  auto &TLI = AM->getResult<TargetLibraryAnalysis>(F);
+  auto *DT = AM->getCachedResult<DominatorTreeAnalysis>(F);
+  return MemoryDependenceResults(AA, AC, TLI, DT);
+}
+
+char MemoryDependenceWrapperPass::ID = 0;
+
+INITIALIZE_PASS_BEGIN(MemoryDependenceWrapperPass, "memdep",
+                      "Memory Dependence Analysis", false, true)
+INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker)
+INITIALIZE_PASS_DEPENDENCY(AAResultsWrapperPass)
+INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)
+INITIALIZE_PASS_END(MemoryDependenceWrapperPass, "memdep",
+                    "Memory Dependence Analysis", false, true)
+
+MemoryDependenceWrapperPass::MemoryDependenceWrapperPass() : FunctionPass(ID) {
+  initializeMemoryDependenceWrapperPassPass(*PassRegistry::getPassRegistry());
+}
+MemoryDependenceWrapperPass::~MemoryDependenceWrapperPass() {}
+
+void MemoryDependenceWrapperPass::releaseMemory() {
+  MemDep.reset();
+}
+
+void MemoryDependenceWrapperPass::getAnalysisUsage(AnalysisUsage &AU) const {
+  AU.setPreservesAll();
+  AU.addRequired<AssumptionCacheTracker>();
+  AU.addRequiredTransitive<AAResultsWrapperPass>();
+  AU.addRequiredTransitive<TargetLibraryInfoWrapperPass>();
+}
+
+bool MemoryDependenceWrapperPass::runOnFunction(Function &F) {
+  auto &AA = getAnalysis<AAResultsWrapperPass>().getAAResults();
+  auto &AC = getAnalysis<AssumptionCacheTracker>().getAssumptionCache(F);
+  auto &TLI = getAnalysis<TargetLibraryInfoWrapperPass>().getTLI();
+  auto *DTWP = getAnalysisIfAvailable<DominatorTreeWrapperPass>();
+  MemDep.emplace(AA, AC, TLI, DTWP ? &DTWP->getDomTree() : nullptr);
+  return false;
+}
+

lib/CodeGen/MachineFunctionPass.cpp

void MachineFunctionPass::getAnalysisUsage(AnalysisUsage &AU) const {
   // CFG in addition to the LLVM IR CFG.
   AU.addPreserved<BasicAAWrapperPass>();
   AU.addPreserved<DominanceFrontierWrapperPass>();
   AU.addPreserved<DominatorTreeWrapperPass>();
   AU.addPreserved<AAResultsWrapperPass>();
   AU.addPreserved<GlobalsAAWrapperPass>();
   AU.addPreserved<IVUsers>();
   AU.addPreserved<LoopInfoWrapperPass>();
-  AU.addPreserved<MemoryDependenceAnalysis>();
+  AU.addPreserved<MemoryDependenceWrapperPass>();
   AU.addPreserved<ScalarEvolutionWrapperPass>();
   AU.addPreserved<SCEVAAWrapperPass>();
   AU.addPreserved<StackProtector>();
 
   FunctionPass::getAnalysisUsage(AU);
 }

lib/Passes/PassBuilder.cpp

 #include "llvm/Analysis/AliasAnalysisEvaluator.h"
 #include "llvm/Analysis/AssumptionCache.h"
 #include "llvm/Analysis/BasicAliasAnalysis.h"
 #include "llvm/Analysis/CFLAliasAnalysis.h"
 #include "llvm/Analysis/CGSCCPassManager.h"
 #include "llvm/Analysis/DominanceFrontier.h"
 #include "llvm/Analysis/LazyCallGraph.h"
 #include "llvm/Analysis/LoopInfo.h"
+#include "llvm/Analysis/MemoryDependenceAnalysis.h"
 #include "llvm/Analysis/PostDominators.h"
 #include "llvm/Analysis/RegionInfo.h"
 #include "llvm/Analysis/ScalarEvolution.h"
 #include "llvm/Analysis/ScalarEvolutionAliasAnalysis.h"
 #include "llvm/Analysis/ScopedNoAliasAA.h"
 #include "llvm/Analysis/TargetLibraryInfo.h"
 #include "llvm/Analysis/TargetTransformInfo.h"
 #include "llvm/Analysis/TypeBasedAliasAnalysis.h"

lib/Passes/PassRegistry.def

 #define FUNCTION_ANALYSIS(NAME, CREATE_PASS)
 #endif
 FUNCTION_ANALYSIS("aa", AAManager())
 FUNCTION_ANALYSIS("assumptions", AssumptionAnalysis())
 FUNCTION_ANALYSIS("domtree", DominatorTreeAnalysis())
 FUNCTION_ANALYSIS("postdomtree", PostDominatorTreeAnalysis())
 FUNCTION_ANALYSIS("domfrontier", DominanceFrontierAnalysis())
 FUNCTION_ANALYSIS("loops", LoopAnalysis())
+FUNCTION_ANALYSIS("memdep", MemoryDependenceAnalysis())
 FUNCTION_ANALYSIS("regions", RegionInfoAnalysis())
 FUNCTION_ANALYSIS("no-op-function", NoOpFunctionAnalysis())
 FUNCTION_ANALYSIS("scalar-evolution", ScalarEvolutionAnalysis())
 FUNCTION_ANALYSIS("targetlibinfo", TargetLibraryAnalysis())
 FUNCTION_ANALYSIS("targetir",
                   TM ? TM->getTargetIRAnalysis() : TargetIRAnalysis())
 
 #ifndef FUNCTION_ALIAS_ANALYSIS

lib/Transforms/Scalar/DeadStoreElimination.cpp

 
 STATISTIC(NumRedundantStores, "Number of redundant stores deleted");
 STATISTIC(NumFastStores, "Number of stores deleted");
 STATISTIC(NumFastOther , "Number of other instrs removed");
 
 namespace {
   struct DSE : public FunctionPass {
     AliasAnalysis *AA;
-    MemoryDependenceAnalysis *MD;
+    MemoryDependenceResults *MD;
     DominatorTree *DT;
     const TargetLibraryInfo *TLI;
 
     static char ID; // Pass identification, replacement for typeid
     DSE() : FunctionPass(ID), AA(nullptr), MD(nullptr), DT(nullptr) {
       initializeDSEPass(*PassRegistry::getPassRegistry());
     }
 
     bool runOnFunction(Function &F) override {
       if (skipOptnoneFunction(F))
         return false;
 
       AA = &getAnalysis<AAResultsWrapperPass>().getAAResults();
-      MD = &getAnalysis<MemoryDependenceAnalysis>();
+      MD = &getAnalysis<MemoryDependenceWrapperPass>().getMemDep();
       DT = &getAnalysis<DominatorTreeWrapperPass>().getDomTree();
       TLI = &getAnalysis<TargetLibraryInfoWrapperPass>().getTLI();
 
       bool Changed = false;
       for (BasicBlock &I : F)
         // Only check non-dead blocks.  Dead blocks may have strange pointer
         // cycles that will confuse alias analysis.
         if (DT->isReachableFromEntry(&I))
struct DSE : public FunctionPass {
     void RemoveAccessedObjects(const MemoryLocation &LoadedLoc,
                                SmallSetVector<Value *, 16> &DeadStackObjects,
                                const DataLayout &DL);
 
     void getAnalysisUsage(AnalysisUsage &AU) const override {
       AU.setPreservesCFG();
       AU.addRequired<DominatorTreeWrapperPass>();
       AU.addRequired<AAResultsWrapperPass>();
-      AU.addRequired<MemoryDependenceAnalysis>();
+      AU.addRequired<MemoryDependenceWrapperPass>();
       AU.addRequired<TargetLibraryInfoWrapperPass>();
       AU.addPreserved<DominatorTreeWrapperPass>();
       AU.addPreserved<GlobalsAAWrapperPass>();
-      AU.addPreserved<MemoryDependenceAnalysis>();
+      AU.addPreserved<MemoryDependenceWrapperPass>();
     }
   };
 }
 
 char DSE::ID = 0;
 INITIALIZE_PASS_BEGIN(DSE, "dse", "Dead Store Elimination", false, false)
 INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(AAResultsWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(GlobalsAAWrapperPass)
-INITIALIZE_PASS_DEPENDENCY(MemoryDependenceAnalysis)
+INITIALIZE_PASS_DEPENDENCY(MemoryDependenceWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)
 INITIALIZE_PASS_END(DSE, "dse", "Dead Store Elimination", false, false)
 
 FunctionPass *llvm::createDeadStoreEliminationPass() { return new DSE(); }
 
 //===----------------------------------------------------------------------===//
 // Helper functions
 //===----------------------------------------------------------------------===//
 
 /// DeleteDeadInstruction - Delete this instruction.  Before we do, go through
 /// and zero out all the operands of this instruction.  If any of them become
 /// dead, delete them and the computation tree that feeds them.
 ///
 /// If ValueSet is non-null, remove any deleted instructions from it as well.
 ///
 static void DeleteDeadInstruction(Instruction *I,
-                               MemoryDependenceAnalysis &MD,
+                               MemoryDependenceResults &MD,
                                const TargetLibraryInfo &TLI,
                                SmallSetVector<Value*, 16> *ValueSet = nullptr) {
   SmallVector<Instruction*, 32> NowDeadInsts;
 
   NowDeadInsts.push_back(I);
   --NumFastOther;
 
   // Before we touch this instruction, remove it from memdep!

lib/Transforms/Scalar/GVN.cpp

friend hash_code hash_value(const Expression &Value) {
                                              Value.varargs.end()));
     }
   };
 
   class ValueTable {
     DenseMap<Value*, uint32_t> valueNumbering;
     DenseMap<Expression, uint32_t> expressionNumbering;
     AliasAnalysis *AA;
-    MemoryDependenceAnalysis *MD;
+    MemoryDependenceResults *MD;
     DominatorTree *DT;
 
     uint32_t nextValueNumber;
 
     Expression create_expression(Instruction* I);
     Expression create_cmp_expression(unsigned Opcode,
                                      CmpInst::Predicate Predicate,
                                      Value *LHS, Value *RHS);
     Expression create_extractvalue_expression(ExtractValueInst* EI);
     uint32_t lookup_or_add_call(CallInst* C);
   public:
     ValueTable() : nextValueNumber(1) { }
     uint32_t lookup_or_add(Value *V);
     uint32_t lookup(Value *V) const;
     uint32_t lookup_or_add_cmp(unsigned Opcode, CmpInst::Predicate Pred,
                                Value *LHS, Value *RHS);
     bool exists(Value *V) const;
     void add(Value *V, uint32_t num);
     void clear();
     void erase(Value *v);
     void setAliasAnalysis(AliasAnalysis* A) { AA = A; }
     AliasAnalysis *getAliasAnalysis() const { return AA; }
-    void setMemDep(MemoryDependenceAnalysis* M) { MD = M; }
+    void setMemDep(MemoryDependenceResults* M) { MD = M; }
     void setDomTree(DominatorTree* D) { DT = D; }
     uint32_t getNextUnusedValueNumber() { return nextValueNumber; }
     void verifyRemoved(const Value *) const;
   };
 }
 
 namespace llvm {
 template <> struct DenseMapInfo<Expression> {
if (local_dep.isDef()) {
       }
 
       uint32_t v = lookup_or_add(local_cdep);
       valueNumbering[C] = v;
       return v;
     }
 
     // Non-local case.
-    const MemoryDependenceAnalysis::NonLocalDepInfo &deps =
+    const MemoryDependenceResults::NonLocalDepInfo &deps =
       MD->getNonLocalCallDependency(CallSite(C));
     // FIXME: Move the checking logic to MemDep!
     CallInst* cdep = nullptr;
 
     // Check to see if we have a single dominating call instruction that is
     // identical to C.
     for (unsigned i = 0, e = deps.size(); i != e; ++i) {
       const NonLocalDepEntry *I = &deps[i];
struct AvailableValueInBlock {
     /// the specified type. This handles various coercion cases.
     Value *MaterializeAdjustedValue(LoadInst *LI, GVN &gvn) const {
       return AV.MaterializeAdjustedValue(LI, BB->getTerminator(), gvn);
     }
   };
 
   class GVN : public FunctionPass {
     bool NoLoads;
-    MemoryDependenceAnalysis *MD;
+    MemoryDependenceResults *MD;
     DominatorTree *DT;
     const TargetLibraryInfo *TLI;
     AssumptionCache *AC;
     SetVector<BasicBlock *> DeadBlocks;
 
     ValueTable VN;
 
     /// A mapping from value numbers to lists of Value*'s that
public:
     /// our various maps and marks it for deletion.
     void markInstructionForDeletion(Instruction *I) {
       VN.erase(I);
       InstrsToErase.push_back(I);
     }
 
     DominatorTree &getDominatorTree() const { return *DT; }
     AliasAnalysis *getAliasAnalysis() const { return VN.getAliasAnalysis(); }
-    MemoryDependenceAnalysis &getMemDep() const { return *MD; }
+    MemoryDependenceResults &getMemDep() const { return *MD; }
   private:
     /// Push a new Value to the LeaderTable onto the list for its value number.
     void addToLeaderTable(uint32_t N, Value *V, const BasicBlock *BB) {
       LeaderTableEntry &Curr = LeaderTable[N];
       if (!Curr.Val) {
         Curr.Val = V;
         Curr.BB = BB;
         return;
private:
     SmallVector<std::pair<TerminatorInst*, unsigned>, 4> toSplit;
 
     // This transformation requires dominator postdominator info
     void getAnalysisUsage(AnalysisUsage &AU) const override {
       AU.addRequired<AssumptionCacheTracker>();
       AU.addRequired<DominatorTreeWrapperPass>();
       AU.addRequired<TargetLibraryInfoWrapperPass>();
       if (!NoLoads)
-        AU.addRequired<MemoryDependenceAnalysis>();
+        AU.addRequired<MemoryDependenceWrapperPass>();
       AU.addRequired<AAResultsWrapperPass>();
 
       AU.addPreserved<DominatorTreeWrapperPass>();
       AU.addPreserved<GlobalsAAWrapperPass>();
     }
 
 
     // Helper functions of redundant load elimination

 
 // The public interface to this file...
 FunctionPass *llvm::createGVNPass(bool NoLoads) {
   return new GVN(NoLoads);
 }
 
 INITIALIZE_PASS_BEGIN(GVN, "gvn", "Global Value Numbering", false, false)
 INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker)
-INITIALIZE_PASS_DEPENDENCY(MemoryDependenceAnalysis)
+INITIALIZE_PASS_DEPENDENCY(MemoryDependenceWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(AAResultsWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(GlobalsAAWrapperPass)
 INITIALIZE_PASS_END(GVN, "gvn", "Global Value Numbering", false, false)
 
 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
 void GVN::dump(DenseMap<uint32_t, Value*>& d) {
static int AnalyzeLoadFromClobberingLoad(Type *LoadTy, Value *LoadPtr,
 
   // If we have a load/load clobber an DepLI can be widened to cover this load,
   // then we should widen it!
   int64_t LoadOffs = 0;
   const Value *LoadBase =
       GetPointerBaseWithConstantOffset(LoadPtr, LoadOffs, DL);
   unsigned LoadSize = DL.getTypeStoreSize(LoadTy);
 
-  unsigned Size = MemoryDependenceAnalysis::getLoadLoadClobberFullWidthSize(
+  unsigned Size = MemoryDependenceResults::getLoadLoadClobberFullWidthSize(
       LoadBase, LoadOffs, LoadSize, DepLI);
   if (Size == 0) return -1;
 
   // Check non-obvious conditions enforced by MDA which we rely on for being
   // able to materialize this potentially available value
   assert(DepLI->isSimple() && "Cannot widen volatile/atomic load!");
   assert(DepLI->getType()->isIntegerTy() && "Can't widen non-integer load");
 

 }
 
 /// runOnFunction - This is the main transformation entry point for a function.
 bool GVN::runOnFunction(Function& F) {
   if (skipOptnoneFunction(F))
     return false;
 
   if (!NoLoads)
-    MD = &getAnalysis<MemoryDependenceAnalysis>();
+    MD = &getAnalysis<MemoryDependenceWrapperPass>().getMemDep();
   DT = &getAnalysis<DominatorTreeWrapperPass>().getDomTree();
   AC = &getAnalysis<AssumptionCacheTracker>().getAssumptionCache(F);
   TLI = &getAnalysis<TargetLibraryInfoWrapperPass>().getTLI();
   VN.setAliasAnalysis(&getAnalysis<AAResultsWrapperPass>().getAAResults());
   VN.setMemDep(MD);
   VN.setDomTree(DT);
 
   bool Changed = false;

lib/Transforms/Scalar/MemCpyOptimizer.cpp

 }
 
 //===----------------------------------------------------------------------===//
 //                         MemCpyOpt Pass
 //===----------------------------------------------------------------------===//
 
 namespace {
   class MemCpyOpt : public FunctionPass {
-    MemoryDependenceAnalysis *MD;
+    MemoryDependenceResults *MD;
     TargetLibraryInfo *TLI;
   public:
     static char ID; // Pass identification, replacement for typeid
     MemCpyOpt() : FunctionPass(ID) {
       initializeMemCpyOptPass(*PassRegistry::getPassRegistry());
       MD = nullptr;
       TLI = nullptr;
     }
 
     bool runOnFunction(Function &F) override;
 
   private:
     // This transformation requires dominator postdominator info
     void getAnalysisUsage(AnalysisUsage &AU) const override {
       AU.setPreservesCFG();
       AU.addRequired<AssumptionCacheTracker>();
       AU.addRequired<DominatorTreeWrapperPass>();
-      AU.addRequired<MemoryDependenceAnalysis>();
+      AU.addRequired<MemoryDependenceWrapperPass>();
       AU.addRequired<AAResultsWrapperPass>();
       AU.addRequired<TargetLibraryInfoWrapperPass>();
       AU.addPreserved<GlobalsAAWrapperPass>();
-      AU.addPreserved<MemoryDependenceAnalysis>();
+      AU.addPreserved<MemoryDependenceWrapperPass>();
     }
 
     // Helper functions
     bool processStore(StoreInst *SI, BasicBlock::iterator &BBI);
     bool processMemSet(MemSetInst *SI, BasicBlock::iterator &BBI);
     bool processMemCpy(MemCpyInst *M);
     bool processMemMove(MemMoveInst *M);
     bool performCallSlotOptzn(Instruction *cpy, Value *cpyDst, Value *cpySrc,

 
 /// The public interface to this file...
 FunctionPass *llvm::createMemCpyOptPass() { return new MemCpyOpt(); }
 
 INITIALIZE_PASS_BEGIN(MemCpyOpt, "memcpyopt", "MemCpy Optimization",
                       false, false)
 INITIALIZE_PASS_DEPENDENCY(AssumptionCacheTracker)
 INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
-INITIALIZE_PASS_DEPENDENCY(MemoryDependenceAnalysis)
+INITIALIZE_PASS_DEPENDENCY(MemoryDependenceWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(AAResultsWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(GlobalsAAWrapperPass)
 INITIALIZE_PASS_END(MemCpyOpt, "memcpyopt", "MemCpy Optimization",
                     false, false)
 
 /// When scanning forward over instructions, we look for some other patterns to
 /// fold away. In particular, this looks for stores to neighboring locations of

 }
 
 /// This is the main transformation entry point for a function.
 bool MemCpyOpt::runOnFunction(Function &F) {
   if (skipOptnoneFunction(F))
     return false;
 
   bool MadeChange = false;
-  MD = &getAnalysis<MemoryDependenceAnalysis>();
+  MD = &getAnalysis<MemoryDependenceWrapperPass>().getMemDep();
   TLI = &getAnalysis<TargetLibraryInfoWrapperPass>().getTLI();
 
   // If we don't have at least memset and memcpy, there is little point of doing
   // anything here.  These are required by a freestanding implementation, so if
   // even they are disabled, there is no point in trying hard.
   if (!TLI->has(LibFunc::memset) || !TLI->has(LibFunc::memcpy))
     return false;
 

lib/Transforms/Scalar/MergedLoadStoreMotion.cpp

 
 //===----------------------------------------------------------------------===//
 //                         MergedLoadStoreMotion Pass
 //===----------------------------------------------------------------------===//
 
 namespace {
 class MergedLoadStoreMotion : public FunctionPass {
   AliasAnalysis *AA;
-  MemoryDependenceAnalysis *MD;
+  MemoryDependenceResults *MD;
 
 public:
   static char ID; // Pass identification, replacement for typeid
   MergedLoadStoreMotion()
       : FunctionPass(ID), MD(nullptr), MagicCompileTimeControl(250) {
     initializeMergedLoadStoreMotionPass(*PassRegistry::getPassRegistry());
   }
 
   bool runOnFunction(Function &F) override;
 
 private:
   // This transformation requires dominator postdominator info
   void getAnalysisUsage(AnalysisUsage &AU) const override {
     AU.setPreservesCFG();
     AU.addRequired<TargetLibraryInfoWrapperPass>();
     AU.addRequired<AAResultsWrapperPass>();
     AU.addPreserved<GlobalsAAWrapperPass>();
-    AU.addPreserved<MemoryDependenceAnalysis>();
+    AU.addPreserved<MemoryDependenceWrapperPass>();
   }
 
   // Helper routines
 
   ///
   /// \brief Remove instruction from parent and update memory dependence
   /// analysis.
   ///

 /// \brief createMergedLoadStoreMotionPass - The public interface to this file.
 ///
 FunctionPass *llvm::createMergedLoadStoreMotionPass() {
   return new MergedLoadStoreMotion();
 }
 
 INITIALIZE_PASS_BEGIN(MergedLoadStoreMotion, "mldst-motion",
                       "MergedLoadStoreMotion", false, false)
-INITIALIZE_PASS_DEPENDENCY(MemoryDependenceAnalysis)
+INITIALIZE_PASS_DEPENDENCY(MemoryDependenceWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfoWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(AAResultsWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(GlobalsAAWrapperPass)
 INITIALIZE_PASS_END(MergedLoadStoreMotion, "mldst-motion",
                     "MergedLoadStoreMotion", false, false)
 
 ///
 /// \brief Remove instruction from parent and update memory dependence analysis.
bool MergedLoadStoreMotion::mergeStores(BasicBlock *T) {
   }
   return MergedStores;
 }
 
 ///
 /// \brief Run the transformation for each function
 ///
 bool MergedLoadStoreMotion::runOnFunction(Function &F) {
-  MD = getAnalysisIfAvailable<MemoryDependenceAnalysis>();
+  auto *MDWP = getAnalysisIfAvailable<MemoryDependenceWrapperPass>();
+  MD = MDWP ? &MDWP->getMemDep() : nullptr;
   AA = &getAnalysis<AAResultsWrapperPass>().getAAResults();
 
   bool Changed = false;
   DEBUG(dbgs() << "Instruction Merger\n");
 
   // Merge unconditional branches, allowing PRE to catch more
   // optimization opportunities.
   for (Function::iterator FI = F.begin(), FE = F.end(); FI != FE;) {

lib/Transforms/Utils/BasicBlockUtils.cpp

void llvm::DeleteDeadBlock(BasicBlock *BB) {
   BB->eraseFromParent();
 }
 
 /// FoldSingleEntryPHINodes - We know that BB has one predecessor.  If there are
 /// any single-entry PHI nodes in it, fold them away.  This handles the case
 /// when all entries to the PHI nodes in a block are guaranteed equal, such as
 /// when the block has exactly one predecessor.
 void llvm::FoldSingleEntryPHINodes(BasicBlock *BB,
-                                   MemoryDependenceAnalysis *MemDep) {
+                                   MemoryDependenceResults *MemDep) {
   if (!isa<PHINode>(BB->begin())) return;
 
   while (PHINode *PN = dyn_cast<PHINode>(BB->begin())) {
     if (PN->getIncomingValue(0) != PN)
       PN->replaceAllUsesWith(PN->getIncomingValue(0));
     else
       PN->replaceAllUsesWith(UndefValue::get(PN->getType()));
 
bool llvm::DeleteDeadPHIs(BasicBlock *BB, const TargetLibraryInfo *TLI) {
 
   return Changed;
 }
 
 /// MergeBlockIntoPredecessor - Attempts to merge a block into its predecessor,
 /// if possible.  The return value indicates success or failure.
 bool llvm::MergeBlockIntoPredecessor(BasicBlock *BB, DominatorTree *DT,
                                      LoopInfo *LI,
-                                     MemoryDependenceAnalysis *MemDep) {
+                                     MemoryDependenceResults *MemDep) {
   // Don't merge away blocks who have their address taken.
   if (BB->hasAddressTaken()) return false;
 
   // Can't merge if there are multiple predecessors, or no predecessors.
   BasicBlock *PredBB = BB->getUniquePredecessor();
   if (!PredBB) return false;
 
   // Don't break self-loops.

test/Other/new-pass-manager.ll

 ; RUN:     -passes='require<aa>' -aa-pipeline='basic-aa' \
 ; RUN:     | FileCheck %s --check-prefix=CHECK-AA
 ; CHECK-AA: Starting llvm::Module pass manager run
 ; CHECK-AA: Running pass: RequireAnalysisPass
 ; CHECK-AA: Running analysis: AAManager
 ; CHECK-AA: Running analysis: BasicAA
 ; CHECK-AA: Finished llvm::Module pass manager run
 
+; RUN: opt -disable-output -disable-verify -debug-pass-manager %s 2>&1 \
+; RUN:     -passes='require<memdep>' \
+; RUN:     | FileCheck %s --check-prefix=CHECK-MEMDEP
+; CHECK-MEMDEP: Starting llvm::Module pass manager run
+; CHECK-MEMDEP: Running pass: RequireAnalysisPass
+; CHECK-MEMDEP: Running analysis: MemoryDependenceAnalysis
+; CHECK-MEMDEP: Finished llvm::Module pass manager run
+
 ; RUN: opt -disable-output -disable-verify -debug-pass-manager \
 ; RUN:     -passes='default<O0>' %s 2>&1 \
 ; RUN:     | FileCheck %s --check-prefix=CHECK-O2
 ; RUN: opt -disable-output -disable-verify -debug-pass-manager \
 ; RUN:     -passes='default<O1>' %s 2>&1 \
 ; RUN:     | FileCheck %s --check-prefix=CHECK-O2
 ; RUN: opt -disable-output -disable-verify -debug-pass-manager \
 ; RUN:     -passes='default<O2>' %s 2>&1 \