- //===- Attributor.h --- Module-wide attribute deduction ---------*- C++ -*-===// 
- // 
- // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 
- // See https://llvm.org/LICENSE.txt for license information. 
- // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 
- // 
- //===----------------------------------------------------------------------===// 
- // 
- // Attributor: An inter procedural (abstract) "attribute" deduction framework. 
- // 
- // The Attributor framework is an inter procedural abstract analysis (fixpoint 
- // iteration analysis). The goal is to allow easy deduction of new attributes as 
- // well as information exchange between abstract attributes in-flight. 
- // 
- // The Attributor class is the driver and the link between the various abstract 
- // attributes. The Attributor will iterate until a fixpoint state is reached by 
- // all abstract attributes in-flight, or until it will enforce a pessimistic fix 
- // point because an iteration limit is reached. 
- // 
- // Abstract attributes, derived from the AbstractAttribute class, actually 
- // describe properties of the code. They can correspond to actual LLVM-IR 
- // attributes, or they can be more general, ultimately unrelated to LLVM-IR 
- // attributes. The latter is useful when an abstract attributes provides 
- // information to other abstract attributes in-flight but we might not want to 
- // manifest the information. The Attributor allows to query in-flight abstract 
- // attributes through the `Attributor::getAAFor` method (see the method 
- // description for an example). If the method is used by an abstract attribute 
- // P, and it results in an abstract attribute Q, the Attributor will 
- // automatically capture a potential dependence from Q to P. This dependence 
- // will cause P to be reevaluated whenever Q changes in the future. 
- // 
- // The Attributor will only reevaluate abstract attributes that might have 
- // changed since the last iteration. That means that the Attribute will not 
- // revisit all instructions/blocks/functions in the module but only query 
- // an update from a subset of the abstract attributes. 
- // 
- // The update method `AbstractAttribute::updateImpl` is implemented by the 
- // specific "abstract attribute" subclasses. The method is invoked whenever the 
- // currently assumed state (see the AbstractState class) might not be valid 
- // anymore. This can, for example, happen if the state was dependent on another 
- // abstract attribute that changed. In every invocation, the update method has 
- // to adjust the internal state of an abstract attribute to a point that is 
- // justifiable by the underlying IR and the current state of abstract attributes 
- // in-flight. Since the IR is given and assumed to be valid, the information 
- // derived from it can be assumed to hold. However, information derived from 
- // other abstract attributes is conditional on various things. If the justifying 
- // state changed, the `updateImpl` has to revisit the situation and potentially 
- // find another justification or limit the optimistic assumes made. 
- // 
- // Change is the key in this framework. Until a state of no-change, thus a 
- // fixpoint, is reached, the Attributor will query the abstract attributes 
- // in-flight to re-evaluate their state. If the (current) state is too 
- // optimistic, hence it cannot be justified anymore through other abstract 
- // attributes or the state of the IR, the state of the abstract attribute will 
- // have to change. Generally, we assume abstract attribute state to be a finite 
- // height lattice and the update function to be monotone. However, these 
- // conditions are not enforced because the iteration limit will guarantee 
- // termination. If an optimistic fixpoint is reached, or a pessimistic fix 
- // point is enforced after a timeout, the abstract attributes are tasked to 
- // manifest their result in the IR for passes to come. 
- // 
- // Attribute manifestation is not mandatory. If desired, there is support to 
- // generate a single or multiple LLVM-IR attributes already in the helper struct 
- // IRAttribute. In the simplest case, a subclass inherits from IRAttribute with 
- // a proper Attribute::AttrKind as template parameter. The Attributor 
- // manifestation framework will then create and place a new attribute if it is 
- // allowed to do so (based on the abstract state). Other use cases can be 
- // achieved by overloading AbstractAttribute or IRAttribute methods. 
- // 
- // 
- // The "mechanics" of adding a new "abstract attribute": 
- // - Define a class (transitively) inheriting from AbstractAttribute and one 
- //   (which could be the same) that (transitively) inherits from AbstractState. 
- //   For the latter, consider the already available BooleanState and 
- //   {Inc,Dec,Bit}IntegerState if they fit your needs, e.g., you require only a 
- //   number tracking or bit-encoding. 
- // - Implement all pure methods. Also use overloading if the attribute is not 
- //   conforming with the "default" behavior: A (set of) LLVM-IR attribute(s) for 
- //   an argument, call site argument, function return value, or function. See 
- //   the class and method descriptions for more information on the two 
- //   "Abstract" classes and their respective methods. 
- // - Register opportunities for the new abstract attribute in the 
- //   `Attributor::identifyDefaultAbstractAttributes` method if it should be 
- //   counted as a 'default' attribute. 
- // - Add sufficient tests. 
- // - Add a Statistics object for bookkeeping. If it is a simple (set of) 
- //   attribute(s) manifested through the Attributor manifestation framework, see 
- //   the bookkeeping function in Attributor.cpp. 
- // - If instructions with a certain opcode are interesting to the attribute, add 
- //   that opcode to the switch in `Attributor::identifyAbstractAttributes`. This 
- //   will make it possible to query all those instructions through the 
- //   `InformationCache::getOpcodeInstMapForFunction` interface and eliminate the 
- //   need to traverse the IR repeatedly. 
- // 
- //===----------------------------------------------------------------------===// 
-   
- #ifndef LLVM_TRANSFORMS_IPO_ATTRIBUTOR_H 
- #define LLVM_TRANSFORMS_IPO_ATTRIBUTOR_H 
-   
- #include "llvm/ADT/DenseSet.h" 
- #include "llvm/ADT/GraphTraits.h" 
- #include "llvm/ADT/MapVector.h" 
- #include "llvm/ADT/STLExtras.h" 
- #include "llvm/ADT/SetOperations.h" 
- #include "llvm/ADT/SetVector.h" 
- #include "llvm/ADT/Triple.h" 
- #include "llvm/ADT/iterator.h" 
- #include "llvm/Analysis/AssumeBundleQueries.h" 
- #include "llvm/Analysis/CFG.h" 
- #include "llvm/Analysis/CGSCCPassManager.h" 
- #include "llvm/Analysis/LazyCallGraph.h" 
- #include "llvm/Analysis/LoopInfo.h" 
- #include "llvm/Analysis/MemoryLocation.h" 
- #include "llvm/Analysis/MustExecute.h" 
- #include "llvm/Analysis/OptimizationRemarkEmitter.h" 
- #include "llvm/Analysis/PostDominators.h" 
- #include "llvm/Analysis/TargetLibraryInfo.h" 
- #include "llvm/IR/AbstractCallSite.h" 
- #include "llvm/IR/ConstantRange.h" 
- #include "llvm/IR/Constants.h" 
- #include "llvm/IR/InstIterator.h" 
- #include "llvm/IR/Instruction.h" 
- #include "llvm/IR/PassManager.h" 
- #include "llvm/IR/Value.h" 
- #include "llvm/Support/Alignment.h" 
- #include "llvm/Support/Allocator.h" 
- #include "llvm/Support/Casting.h" 
- #include "llvm/Support/DOTGraphTraits.h" 
- #include "llvm/Support/TimeProfiler.h" 
- #include "llvm/Transforms/Utils/CallGraphUpdater.h" 
-   
- #include <limits> 
- #include <map> 
- #include <optional> 
-   
- namespace llvm { 
-   
- class DataLayout; 
- class LLVMContext; 
- class Pass; 
- template <typename Fn> class function_ref; 
- struct AADepGraphNode; 
- struct AADepGraph; 
- struct Attributor; 
- struct AbstractAttribute; 
- struct InformationCache; 
- struct AAIsDead; 
- struct AttributorCallGraph; 
- struct IRPosition; 
-   
- class AAResults; 
- class Function; 
-   
- /// Abstract Attribute helper functions. 
- namespace AA { 
- using InstExclusionSetTy = SmallPtrSet<Instruction *, 4>; 
-   
- enum class GPUAddressSpace : unsigned { 
-   Generic = 0, 
-   Global = 1, 
-   Shared = 3, 
-   Constant = 4, 
-   Local = 5, 
- }; 
-   
- /// Flags to distinguish intra-procedural queries from *potentially* 
- /// inter-procedural queries. Not that information can be valid for both and 
- /// therefore both bits might be set. 
- enum ValueScope : uint8_t { 
-   Intraprocedural = 1, 
-   Interprocedural = 2, 
-   AnyScope = Intraprocedural | Interprocedural, 
- }; 
-   
- struct ValueAndContext : public std::pair<Value *, const Instruction *> { 
-   using Base = std::pair<Value *, const Instruction *>; 
-   ValueAndContext(const Base &B) : Base(B) {} 
-   ValueAndContext(Value &V, const Instruction *CtxI) : Base(&V, CtxI) {} 
-   ValueAndContext(Value &V, const Instruction &CtxI) : Base(&V, &CtxI) {} 
-   
-   Value *getValue() const { return this->first; } 
-   const Instruction *getCtxI() const { return this->second; } 
- }; 
-   
- /// Return true if \p I is a `nosync` instruction. Use generic reasoning and 
- /// potentially the corresponding AANoSync. 
- bool isNoSyncInst(Attributor &A, const Instruction &I, 
-                   const AbstractAttribute &QueryingAA); 
-   
- /// Return true if \p V is dynamically unique, that is, there are no two 
- /// "instances" of \p V at runtime with different values. 
- /// Note: If \p ForAnalysisOnly is set we only check that the Attributor will 
- /// never use \p V to represent two "instances" not that \p V could not 
- /// technically represent them. 
- bool isDynamicallyUnique(Attributor &A, const AbstractAttribute &QueryingAA, 
-                          const Value &V, bool ForAnalysisOnly = true); 
-   
- /// Return true if \p V is a valid value in \p Scope, that is a constant or an 
- /// instruction/argument of \p Scope. 
- bool isValidInScope(const Value &V, const Function *Scope); 
-   
- /// Return true if the value of \p VAC is a valid at the position of \p VAC, 
- /// that is a constant, an argument of the same function, or an instruction in 
- /// that function that dominates the position. 
- bool isValidAtPosition(const ValueAndContext &VAC, InformationCache &InfoCache); 
-   
- /// Try to convert \p V to type \p Ty without introducing new instructions. If 
- /// this is not possible return `nullptr`. Note: this function basically knows 
- /// how to cast various constants. 
- Value *getWithType(Value &V, Type &Ty); 
-   
- /// Return the combination of \p A and \p B such that the result is a possible 
- /// value of both. \p B is potentially casted to match the type \p Ty or the 
- /// type of \p A if \p Ty is null. 
- /// 
- /// Examples: 
- ///        X + none  => X 
- /// not_none + undef => not_none 
- ///          V1 + V2 => nullptr 
- std::optional<Value *> 
- combineOptionalValuesInAAValueLatice(const std::optional<Value *> &A, 
-                                      const std::optional<Value *> &B, Type *Ty); 
-   
- /// Helper to represent an access offset and size, with logic to deal with 
- /// uncertainty and check for overlapping accesses. 
- struct RangeTy { 
-   int64_t Offset = Unassigned; 
-   int64_t Size = Unassigned; 
-   
-   RangeTy(int64_t Offset, int64_t Size) : Offset(Offset), Size(Size) {} 
-   RangeTy() = default; 
-   static RangeTy getUnknown() { return RangeTy{Unknown, Unknown}; } 
-   
-   /// Return true if offset or size are unknown. 
-   bool offsetOrSizeAreUnknown() const { 
-     return Offset == RangeTy::Unknown || Size == RangeTy::Unknown; 
-   } 
-   
-   /// Return true if offset and size are unknown, thus this is the default 
-   /// unknown object. 
-   bool offsetAndSizeAreUnknown() const { 
-     return Offset == RangeTy::Unknown && Size == RangeTy::Unknown; 
-   } 
-   
-   /// Return true if the offset and size are unassigned. 
-   bool isUnassigned() const { 
-     assert((Offset == RangeTy::Unassigned) == (Size == RangeTy::Unassigned) && 
-            "Inconsistent state!"); 
-     return Offset == RangeTy::Unassigned; 
-   } 
-   
-   /// Return true if this offset and size pair might describe an address that 
-   /// overlaps with \p Range. 
-   bool mayOverlap(const RangeTy &Range) const { 
-     // Any unknown value and we are giving up -> overlap. 
-     if (offsetOrSizeAreUnknown() || Range.offsetOrSizeAreUnknown()) 
-       return true; 
-   
-     // Check if one offset point is in the other interval [offset, 
-     // offset+size]. 
-     return Range.Offset + Range.Size > Offset && Range.Offset < Offset + Size; 
-   } 
-   
-   RangeTy &operator&=(const RangeTy &R) { 
-     if (Offset == Unassigned) 
-       Offset = R.Offset; 
-     else if (R.Offset != Unassigned && R.Offset != Offset) 
-       Offset = Unknown; 
-   
-     if (Size == Unassigned) 
-       Size = R.Size; 
-     else if (Size == Unknown || R.Size == Unknown) 
-       Size = Unknown; 
-     else if (R.Size != Unassigned) 
-       Size = std::max(Size, R.Size); 
-   
-     return *this; 
-   } 
-   
-   /// Comparison for sorting ranges by offset. 
-   /// 
-   /// Returns true if the offset \p L is less than that of \p R. 
-   inline static bool OffsetLessThan(const RangeTy &L, const RangeTy &R) { 
-     return L.Offset < R.Offset; 
-   } 
-   
-   /// Constants used to represent special offsets or sizes. 
-   /// - We cannot assume that Offsets and Size are non-negative. 
-   /// - The constants should not clash with DenseMapInfo, such as EmptyKey 
-   ///   (INT64_MAX) and TombstoneKey (INT64_MIN). 
-   /// We use values "in the middle" of the 64 bit range to represent these 
-   /// special cases. 
-   static constexpr int64_t Unassigned = std::numeric_limits<int32_t>::min(); 
-   static constexpr int64_t Unknown = std::numeric_limits<int32_t>::max(); 
- }; 
-   
- inline raw_ostream &operator<<(raw_ostream &OS, const RangeTy &R) { 
-   OS << "[" << R.Offset << ", " << R.Size << "]"; 
-   return OS; 
- } 
-   
- inline bool operator==(const RangeTy &A, const RangeTy &B) { 
-   return A.Offset == B.Offset && A.Size == B.Size; 
- } 
-   
- inline bool operator!=(const RangeTy &A, const RangeTy &B) { return !(A == B); } 
-   
- /// Return the initial value of \p Obj with type \p Ty if that is a constant. 
- Constant *getInitialValueForObj(Value &Obj, Type &Ty, 
-                                 const TargetLibraryInfo *TLI, 
-                                 const DataLayout &DL, 
-                                 RangeTy *RangePtr = nullptr); 
-   
- /// Collect all potential values \p LI could read into \p PotentialValues. That 
- /// is, the only values read by \p LI are assumed to be known and all are in 
- /// \p PotentialValues. \p PotentialValueOrigins will contain all the 
- /// instructions that might have put a potential value into \p PotentialValues. 
- /// Dependences onto \p QueryingAA are properly tracked, \p 
- /// UsedAssumedInformation will inform the caller if assumed information was 
- /// used. 
- /// 
- /// \returns True if the assumed potential copies are all in \p PotentialValues, 
- ///          false if something went wrong and the copies could not be 
- ///          determined. 
- bool getPotentiallyLoadedValues( 
-     Attributor &A, LoadInst &LI, SmallSetVector<Value *, 4> &PotentialValues, 
-     SmallSetVector<Instruction *, 4> &PotentialValueOrigins, 
-     const AbstractAttribute &QueryingAA, bool &UsedAssumedInformation, 
-     bool OnlyExact = false); 
-   
- /// Collect all potential values of the one stored by \p SI into 
- /// \p PotentialCopies. That is, the only copies that were made via the 
- /// store are assumed to be known and all are in \p PotentialCopies. Dependences 
- /// onto \p QueryingAA are properly tracked, \p UsedAssumedInformation will 
- /// inform the caller if assumed information was used. 
- /// 
- /// \returns True if the assumed potential copies are all in \p PotentialCopies, 
- ///          false if something went wrong and the copies could not be 
- ///          determined. 
- bool getPotentialCopiesOfStoredValue( 
-     Attributor &A, StoreInst &SI, SmallSetVector<Value *, 4> &PotentialCopies, 
-     const AbstractAttribute &QueryingAA, bool &UsedAssumedInformation, 
-     bool OnlyExact = false); 
-   
- /// Return true if \p IRP is readonly. This will query respective AAs that 
- /// deduce the information and introduce dependences for \p QueryingAA. 
- bool isAssumedReadOnly(Attributor &A, const IRPosition &IRP, 
-                        const AbstractAttribute &QueryingAA, bool &IsKnown); 
-   
- /// Return true if \p IRP is readnone. This will query respective AAs that 
- /// deduce the information and introduce dependences for \p QueryingAA. 
- bool isAssumedReadNone(Attributor &A, const IRPosition &IRP, 
-                        const AbstractAttribute &QueryingAA, bool &IsKnown); 
-   
- /// Return true if \p ToI is potentially reachable from \p FromI without running 
- /// into any instruction in \p ExclusionSet The two instructions do not need to 
- /// be in the same function. \p GoBackwardsCB can be provided to convey domain 
- /// knowledge about the "lifespan" the user is interested in. By default, the 
- /// callers of \p FromI are checked as well to determine if \p ToI can be 
- /// reached. If the query is not interested in callers beyond a certain point, 
- /// e.g., a GPU kernel entry or the function containing an alloca, the 
- /// \p GoBackwardsCB should return false. 
- bool isPotentiallyReachable( 
-     Attributor &A, const Instruction &FromI, const Instruction &ToI, 
-     const AbstractAttribute &QueryingAA, 
-     const AA::InstExclusionSetTy *ExclusionSet = nullptr, 
-     std::function<bool(const Function &F)> GoBackwardsCB = nullptr); 
-   
- /// Same as above but it is sufficient to reach any instruction in \p ToFn. 
- bool isPotentiallyReachable( 
-     Attributor &A, const Instruction &FromI, const Function &ToFn, 
-     const AbstractAttribute &QueryingAA, 
-     const AA::InstExclusionSetTy *ExclusionSet = nullptr, 
-     std::function<bool(const Function &F)> GoBackwardsCB = nullptr); 
-   
- /// Return true if \p Obj is assumed to be a thread local object. 
- bool isAssumedThreadLocalObject(Attributor &A, Value &Obj, 
-                                 const AbstractAttribute &QueryingAA); 
-   
- /// Return true if \p I is potentially affected by a barrier. 
- bool isPotentiallyAffectedByBarrier(Attributor &A, const Instruction &I, 
-                                     const AbstractAttribute &QueryingAA); 
- bool isPotentiallyAffectedByBarrier(Attributor &A, ArrayRef<const Value *> Ptrs, 
-                                     const AbstractAttribute &QueryingAA, 
-                                     const Instruction *CtxI); 
- } // namespace AA 
-   
- template <> 
- struct DenseMapInfo<AA::ValueAndContext> 
-     : public DenseMapInfo<AA::ValueAndContext::Base> { 
-   using Base = DenseMapInfo<AA::ValueAndContext::Base>; 
-   static inline AA::ValueAndContext getEmptyKey() { 
-     return Base::getEmptyKey(); 
-   } 
-   static inline AA::ValueAndContext getTombstoneKey() { 
-     return Base::getTombstoneKey(); 
-   } 
-   static unsigned getHashValue(const AA::ValueAndContext &VAC) { 
-     return Base::getHashValue(VAC); 
-   } 
-   
-   static bool isEqual(const AA::ValueAndContext &LHS, 
-                       const AA::ValueAndContext &RHS) { 
-     return Base::isEqual(LHS, RHS); 
-   } 
- }; 
-   
- template <> 
- struct DenseMapInfo<AA::ValueScope> : public DenseMapInfo<unsigned char> { 
-   using Base = DenseMapInfo<unsigned char>; 
-   static inline AA::ValueScope getEmptyKey() { 
-     return AA::ValueScope(Base::getEmptyKey()); 
-   } 
-   static inline AA::ValueScope getTombstoneKey() { 
-     return AA::ValueScope(Base::getTombstoneKey()); 
-   } 
-   static unsigned getHashValue(const AA::ValueScope &S) { 
-     return Base::getHashValue(S); 
-   } 
-   
-   static bool isEqual(const AA::ValueScope &LHS, const AA::ValueScope &RHS) { 
-     return Base::isEqual(LHS, RHS); 
-   } 
- }; 
-   
- template <> 
- struct DenseMapInfo<const AA::InstExclusionSetTy *> 
-     : public DenseMapInfo<void *> { 
-   using super = DenseMapInfo<void *>; 
-   static inline const AA::InstExclusionSetTy *getEmptyKey() { 
-     return static_cast<const AA::InstExclusionSetTy *>(super::getEmptyKey()); 
-   } 
-   static inline const AA::InstExclusionSetTy *getTombstoneKey() { 
-     return static_cast<const AA::InstExclusionSetTy *>( 
-         super::getTombstoneKey()); 
-   } 
-   static unsigned getHashValue(const AA::InstExclusionSetTy *BES) { 
-     unsigned H = 0; 
-     if (BES) 
-       for (const auto *II : *BES) 
-         H += DenseMapInfo<const Instruction *>::getHashValue(II); 
-     return H; 
-   } 
-   static bool isEqual(const AA::InstExclusionSetTy *LHS, 
-                       const AA::InstExclusionSetTy *RHS) { 
-     if (LHS == RHS) 
-       return true; 
-     if (LHS == getEmptyKey() || RHS == getEmptyKey() || 
-         LHS == getTombstoneKey() || RHS == getTombstoneKey()) 
-       return false; 
-     if (!LHS || !RHS) 
-       return ((LHS && LHS->empty()) || (RHS && RHS->empty())); 
-     if (LHS->size() != RHS->size()) 
-       return false; 
-     return llvm::set_is_subset(*LHS, *RHS); 
-   } 
- }; 
-   
- /// The value passed to the line option that defines the maximal initialization 
- /// chain length. 
- extern unsigned MaxInitializationChainLength; 
-   
- ///{ 
- enum class ChangeStatus { 
-   CHANGED, 
-   UNCHANGED, 
- }; 
-   
- ChangeStatus operator|(ChangeStatus l, ChangeStatus r); 
- ChangeStatus &operator|=(ChangeStatus &l, ChangeStatus r); 
- ChangeStatus operator&(ChangeStatus l, ChangeStatus r); 
- ChangeStatus &operator&=(ChangeStatus &l, ChangeStatus r); 
-   
- enum class DepClassTy { 
-   REQUIRED, ///< The target cannot be valid if the source is not. 
-   OPTIONAL, ///< The target may be valid if the source is not. 
-   NONE,     ///< Do not track a dependence between source and target. 
- }; 
- ///} 
-   
- /// The data structure for the nodes of a dependency graph 
- struct AADepGraphNode { 
- public: 
-   virtual ~AADepGraphNode() = default; 
-   using DepTy = PointerIntPair<AADepGraphNode *, 1>; 
-   
- protected: 
-   /// Set of dependency graph nodes which should be updated if this one 
-   /// is updated. The bit encodes if it is optional. 
-   TinyPtrVector<DepTy> Deps; 
-   
-   static AADepGraphNode *DepGetVal(DepTy &DT) { return DT.getPointer(); } 
-   static AbstractAttribute *DepGetValAA(DepTy &DT) { 
-     return cast<AbstractAttribute>(DT.getPointer()); 
-   } 
-   
-   operator AbstractAttribute *() { return cast<AbstractAttribute>(this); } 
-   
- public: 
-   using iterator = 
-       mapped_iterator<TinyPtrVector<DepTy>::iterator, decltype(&DepGetVal)>; 
-   using aaiterator = 
-       mapped_iterator<TinyPtrVector<DepTy>::iterator, decltype(&DepGetValAA)>; 
-   
-   aaiterator begin() { return aaiterator(Deps.begin(), &DepGetValAA); } 
-   aaiterator end() { return aaiterator(Deps.end(), &DepGetValAA); } 
-   iterator child_begin() { return iterator(Deps.begin(), &DepGetVal); } 
-   iterator child_end() { return iterator(Deps.end(), &DepGetVal); } 
-   
-   virtual void print(raw_ostream &OS) const { OS << "AADepNode Impl\n"; } 
-   TinyPtrVector<DepTy> &getDeps() { return Deps; } 
-   
-   friend struct Attributor; 
-   friend struct AADepGraph; 
- }; 
-   
- /// The data structure for the dependency graph 
- /// 
- /// Note that in this graph if there is an edge from A to B (A -> B), 
- /// then it means that B depends on A, and when the state of A is 
- /// updated, node B should also be updated 
- struct AADepGraph { 
-   AADepGraph() = default; 
-   ~AADepGraph() = default; 
-   
-   using DepTy = AADepGraphNode::DepTy; 
-   static AADepGraphNode *DepGetVal(DepTy &DT) { return DT.getPointer(); } 
-   using iterator = 
-       mapped_iterator<TinyPtrVector<DepTy>::iterator, decltype(&DepGetVal)>; 
-   
-   /// There is no root node for the dependency graph. But the SCCIterator 
-   /// requires a single entry point, so we maintain a fake("synthetic") root 
-   /// node that depends on every node. 
-   AADepGraphNode SyntheticRoot; 
-   AADepGraphNode *GetEntryNode() { return &SyntheticRoot; } 
-   
-   iterator begin() { return SyntheticRoot.child_begin(); } 
-   iterator end() { return SyntheticRoot.child_end(); } 
-   
-   void viewGraph(); 
-   
-   /// Dump graph to file 
-   void dumpGraph(); 
-   
-   /// Print dependency graph 
-   void print(); 
- }; 
-   
- /// Helper to describe and deal with positions in the LLVM-IR. 
- /// 
- /// A position in the IR is described by an anchor value and an "offset" that 
- /// could be the argument number, for call sites and arguments, or an indicator 
- /// of the "position kind". The kinds, specified in the Kind enum below, include 
- /// the locations in the attribute list, i.a., function scope and return value, 
- /// as well as a distinction between call sites and functions. Finally, there 
- /// are floating values that do not have a corresponding attribute list 
- /// position. 
- struct IRPosition { 
-   // NOTE: In the future this definition can be changed to support recursive 
-   // functions. 
-   using CallBaseContext = CallBase; 
-   
-   /// The positions we distinguish in the IR. 
-   enum Kind : char { 
-     IRP_INVALID,  ///< An invalid position. 
-     IRP_FLOAT,    ///< A position that is not associated with a spot suitable 
-                   ///< for attributes. This could be any value or instruction. 
-     IRP_RETURNED, ///< An attribute for the function return value. 
-     IRP_CALL_SITE_RETURNED, ///< An attribute for a call site return value. 
-     IRP_FUNCTION,           ///< An attribute for a function (scope). 
-     IRP_CALL_SITE,          ///< An attribute for a call site (function scope). 
-     IRP_ARGUMENT,           ///< An attribute for a function argument. 
-     IRP_CALL_SITE_ARGUMENT, ///< An attribute for a call site argument. 
-   }; 
-   
-   /// Default constructor available to create invalid positions implicitly. All 
-   /// other positions need to be created explicitly through the appropriate 
-   /// static member function. 
-   IRPosition() : Enc(nullptr, ENC_VALUE) { verify(); } 
-   
-   /// Create a position describing the value of \p V. 
-   static const IRPosition value(const Value &V, 
-                                 const CallBaseContext *CBContext = nullptr) { 
-     if (auto *Arg = dyn_cast<Argument>(&V)) 
-       return IRPosition::argument(*Arg, CBContext); 
-     if (auto *CB = dyn_cast<CallBase>(&V)) 
-       return IRPosition::callsite_returned(*CB); 
-     return IRPosition(const_cast<Value &>(V), IRP_FLOAT, CBContext); 
-   } 
-   
-   /// Create a position describing the instruction \p I. This is different from 
-   /// the value version because call sites are treated as intrusctions rather 
-   /// than their return value in this function. 
-   static const IRPosition inst(const Instruction &I, 
-                                const CallBaseContext *CBContext = nullptr) { 
-     return IRPosition(const_cast<Instruction &>(I), IRP_FLOAT, CBContext); 
-   } 
-   
-   /// Create a position describing the function scope of \p F. 
-   /// \p CBContext is used for call base specific analysis. 
-   static const IRPosition function(const Function &F, 
-                                    const CallBaseContext *CBContext = nullptr) { 
-     return IRPosition(const_cast<Function &>(F), IRP_FUNCTION, CBContext); 
-   } 
-   
-   /// Create a position describing the returned value of \p F. 
-   /// \p CBContext is used for call base specific analysis. 
-   static const IRPosition returned(const Function &F, 
-                                    const CallBaseContext *CBContext = nullptr) { 
-     return IRPosition(const_cast<Function &>(F), IRP_RETURNED, CBContext); 
-   } 
-   
-   /// Create a position describing the argument \p Arg. 
-   /// \p CBContext is used for call base specific analysis. 
-   static const IRPosition argument(const Argument &Arg, 
-                                    const CallBaseContext *CBContext = nullptr) { 
-     return IRPosition(const_cast<Argument &>(Arg), IRP_ARGUMENT, CBContext); 
-   } 
-   
-   /// Create a position describing the function scope of \p CB. 
-   static const IRPosition callsite_function(const CallBase &CB) { 
-     return IRPosition(const_cast<CallBase &>(CB), IRP_CALL_SITE); 
-   } 
-   
-   /// Create a position describing the returned value of \p CB. 
-   static const IRPosition callsite_returned(const CallBase &CB) { 
-     return IRPosition(const_cast<CallBase &>(CB), IRP_CALL_SITE_RETURNED); 
-   } 
-   
-   /// Create a position describing the argument of \p CB at position \p ArgNo. 
-   static const IRPosition callsite_argument(const CallBase &CB, 
-                                             unsigned ArgNo) { 
-     return IRPosition(const_cast<Use &>(CB.getArgOperandUse(ArgNo)), 
-                       IRP_CALL_SITE_ARGUMENT); 
-   } 
-   
-   /// Create a position describing the argument of \p ACS at position \p ArgNo. 
-   static const IRPosition callsite_argument(AbstractCallSite ACS, 
-                                             unsigned ArgNo) { 
-     if (ACS.getNumArgOperands() <= ArgNo) 
-       return IRPosition(); 
-     int CSArgNo = ACS.getCallArgOperandNo(ArgNo); 
-     if (CSArgNo >= 0) 
-       return IRPosition::callsite_argument( 
-           cast<CallBase>(*ACS.getInstruction()), CSArgNo); 
-     return IRPosition(); 
-   } 
-   
-   /// Create a position with function scope matching the "context" of \p IRP. 
-   /// If \p IRP is a call site (see isAnyCallSitePosition()) then the result 
-   /// will be a call site position, otherwise the function position of the 
-   /// associated function. 
-   static const IRPosition 
-   function_scope(const IRPosition &IRP, 
-                  const CallBaseContext *CBContext = nullptr) { 
-     if (IRP.isAnyCallSitePosition()) { 
-       return IRPosition::callsite_function( 
-           cast<CallBase>(IRP.getAnchorValue())); 
-     } 
-     assert(IRP.getAssociatedFunction()); 
-     return IRPosition::function(*IRP.getAssociatedFunction(), CBContext); 
-   } 
-   
-   bool operator==(const IRPosition &RHS) const { 
-     return Enc == RHS.Enc && RHS.CBContext == CBContext; 
-   } 
-   bool operator!=(const IRPosition &RHS) const { return !(*this == RHS); } 
-   
-   /// Return the value this abstract attribute is anchored with. 
-   /// 
-   /// The anchor value might not be the associated value if the latter is not 
-   /// sufficient to determine where arguments will be manifested. This is, so 
-   /// far, only the case for call site arguments as the value is not sufficient 
-   /// to pinpoint them. Instead, we can use the call site as an anchor. 
-   Value &getAnchorValue() const { 
-     switch (getEncodingBits()) { 
-     case ENC_VALUE: 
-     case ENC_RETURNED_VALUE: 
-     case ENC_FLOATING_FUNCTION: 
-       return *getAsValuePtr(); 
-     case ENC_CALL_SITE_ARGUMENT_USE: 
-       return *(getAsUsePtr()->getUser()); 
-     default: 
-       llvm_unreachable("Unkown encoding!"); 
-     }; 
-   } 
-   
-   /// Return the associated function, if any. 
-   Function *getAssociatedFunction() const { 
-     if (auto *CB = dyn_cast<CallBase>(&getAnchorValue())) { 
-       // We reuse the logic that associates callback calles to arguments of a 
-       // call site here to identify the callback callee as the associated 
-       // function. 
-       if (Argument *Arg = getAssociatedArgument()) 
-         return Arg->getParent(); 
-       return CB->getCalledFunction(); 
-     } 
-     return getAnchorScope(); 
-   } 
-   
-   /// Return the associated argument, if any. 
-   Argument *getAssociatedArgument() const; 
-   
-   /// Return true if the position refers to a function interface, that is the 
-   /// function scope, the function return, or an argument. 
-   bool isFnInterfaceKind() const { 
-     switch (getPositionKind()) { 
-     case IRPosition::IRP_FUNCTION: 
-     case IRPosition::IRP_RETURNED: 
-     case IRPosition::IRP_ARGUMENT: 
-       return true; 
-     default: 
-       return false; 
-     } 
-   } 
-   
-   /// Return the Function surrounding the anchor value. 
-   Function *getAnchorScope() const { 
-     Value &V = getAnchorValue(); 
-     if (isa<Function>(V)) 
-       return &cast<Function>(V); 
-     if (isa<Argument>(V)) 
-       return cast<Argument>(V).getParent(); 
-     if (isa<Instruction>(V)) 
-       return cast<Instruction>(V).getFunction(); 
-     return nullptr; 
-   } 
-   
-   /// Return the context instruction, if any. 
-   Instruction *getCtxI() const { 
-     Value &V = getAnchorValue(); 
-     if (auto *I = dyn_cast<Instruction>(&V)) 
-       return I; 
-     if (auto *Arg = dyn_cast<Argument>(&V)) 
-       if (!Arg->getParent()->isDeclaration()) 
-         return &Arg->getParent()->getEntryBlock().front(); 
-     if (auto *F = dyn_cast<Function>(&V)) 
-       if (!F->isDeclaration()) 
-         return &(F->getEntryBlock().front()); 
-     return nullptr; 
-   } 
-   
-   /// Return the value this abstract attribute is associated with. 
-   Value &getAssociatedValue() const { 
-     if (getCallSiteArgNo() < 0 || isa<Argument>(&getAnchorValue())) 
-       return getAnchorValue(); 
-     assert(isa<CallBase>(&getAnchorValue()) && "Expected a call base!"); 
-     return *cast<CallBase>(&getAnchorValue()) 
-                 ->getArgOperand(getCallSiteArgNo()); 
-   } 
-   
-   /// Return the type this abstract attribute is associated with. 
-   Type *getAssociatedType() const { 
-     if (getPositionKind() == IRPosition::IRP_RETURNED) 
-       return getAssociatedFunction()->getReturnType(); 
-     return getAssociatedValue().getType(); 
-   } 
-   
-   /// Return the callee argument number of the associated value if it is an 
-   /// argument or call site argument, otherwise a negative value. In contrast to 
-   /// `getCallSiteArgNo` this method will always return the "argument number" 
-   /// from the perspective of the callee. This may not the same as the call site 
-   /// if this is a callback call. 
-   int getCalleeArgNo() const { 
-     return getArgNo(/* CallbackCalleeArgIfApplicable */ true); 
-   } 
-   
-   /// Return the call site argument number of the associated value if it is an 
-   /// argument or call site argument, otherwise a negative value. In contrast to 
-   /// `getCalleArgNo` this method will always return the "operand number" from 
-   /// the perspective of the call site. This may not the same as the callee 
-   /// perspective if this is a callback call. 
-   int getCallSiteArgNo() const { 
-     return getArgNo(/* CallbackCalleeArgIfApplicable */ false); 
-   } 
-   
-   /// Return the index in the attribute list for this position. 
-   unsigned getAttrIdx() const { 
-     switch (getPositionKind()) { 
-     case IRPosition::IRP_INVALID: 
-     case IRPosition::IRP_FLOAT: 
-       break; 
-     case IRPosition::IRP_FUNCTION: 
-     case IRPosition::IRP_CALL_SITE: 
-       return AttributeList::FunctionIndex; 
-     case IRPosition::IRP_RETURNED: 
-     case IRPosition::IRP_CALL_SITE_RETURNED: 
-       return AttributeList::ReturnIndex; 
-     case IRPosition::IRP_ARGUMENT: 
-     case IRPosition::IRP_CALL_SITE_ARGUMENT: 
-       return getCallSiteArgNo() + AttributeList::FirstArgIndex; 
-     } 
-     llvm_unreachable( 
-         "There is no attribute index for a floating or invalid position!"); 
-   } 
-   
-   /// Return the associated position kind. 
-   Kind getPositionKind() const { 
-     char EncodingBits = getEncodingBits(); 
-     if (EncodingBits == ENC_CALL_SITE_ARGUMENT_USE) 
-       return IRP_CALL_SITE_ARGUMENT; 
-     if (EncodingBits == ENC_FLOATING_FUNCTION) 
-       return IRP_FLOAT; 
-   
-     Value *V = getAsValuePtr(); 
-     if (!V) 
-       return IRP_INVALID; 
-     if (isa<Argument>(V)) 
-       return IRP_ARGUMENT; 
-     if (isa<Function>(V)) 
-       return isReturnPosition(EncodingBits) ? IRP_RETURNED : IRP_FUNCTION; 
-     if (isa<CallBase>(V)) 
-       return isReturnPosition(EncodingBits) ? IRP_CALL_SITE_RETURNED 
-                                             : IRP_CALL_SITE; 
-     return IRP_FLOAT; 
-   } 
-   
-   /// TODO: Figure out if the attribute related helper functions should live 
-   ///       here or somewhere else. 
-   
-   /// Return true if any kind in \p AKs existing in the IR at a position that 
-   /// will affect this one. See also getAttrs(...). 
-   /// \param IgnoreSubsumingPositions Flag to determine if subsuming positions, 
-   ///                                 e.g., the function position if this is an 
-   ///                                 argument position, should be ignored. 
-   bool hasAttr(ArrayRef<Attribute::AttrKind> AKs, 
-                bool IgnoreSubsumingPositions = false, 
-                Attributor *A = nullptr) const; 
-   
-   /// Return the attributes of any kind in \p AKs existing in the IR at a 
-   /// position that will affect this one. While each position can only have a 
-   /// single attribute of any kind in \p AKs, there are "subsuming" positions 
-   /// that could have an attribute as well. This method returns all attributes 
-   /// found in \p Attrs. 
-   /// \param IgnoreSubsumingPositions Flag to determine if subsuming positions, 
-   ///                                 e.g., the function position if this is an 
-   ///                                 argument position, should be ignored. 
-   void getAttrs(ArrayRef<Attribute::AttrKind> AKs, 
-                 SmallVectorImpl<Attribute> &Attrs, 
-                 bool IgnoreSubsumingPositions = false, 
-                 Attributor *A = nullptr) const; 
-   
-   /// Remove the attribute of kind \p AKs existing in the IR at this position. 
-   void removeAttrs(ArrayRef<Attribute::AttrKind> AKs) const { 
-     if (getPositionKind() == IRP_INVALID || getPositionKind() == IRP_FLOAT) 
-       return; 
-   
-     AttributeList AttrList; 
-     auto *CB = dyn_cast<CallBase>(&getAnchorValue()); 
-     if (CB) 
-       AttrList = CB->getAttributes(); 
-     else 
-       AttrList = getAssociatedFunction()->getAttributes(); 
-   
-     LLVMContext &Ctx = getAnchorValue().getContext(); 
-     for (Attribute::AttrKind AK : AKs) 
-       AttrList = AttrList.removeAttributeAtIndex(Ctx, getAttrIdx(), AK); 
-   
-     if (CB) 
-       CB->setAttributes(AttrList); 
-     else 
-       getAssociatedFunction()->setAttributes(AttrList); 
-   } 
-   
-   bool isAnyCallSitePosition() const { 
-     switch (getPositionKind()) { 
-     case IRPosition::IRP_CALL_SITE: 
-     case IRPosition::IRP_CALL_SITE_RETURNED: 
-     case IRPosition::IRP_CALL_SITE_ARGUMENT: 
-       return true; 
-     default: 
-       return false; 
-     } 
-   } 
-   
-   /// Return true if the position is an argument or call site argument. 
-   bool isArgumentPosition() const { 
-     switch (getPositionKind()) { 
-     case IRPosition::IRP_ARGUMENT: 
-     case IRPosition::IRP_CALL_SITE_ARGUMENT: 
-       return true; 
-     default: 
-       return false; 
-     } 
-   } 
-   
-   /// Return the same position without the call base context. 
-   IRPosition stripCallBaseContext() const { 
-     IRPosition Result = *this; 
-     Result.CBContext = nullptr; 
-     return Result; 
-   } 
-   
-   /// Get the call base context from the position. 
-   const CallBaseContext *getCallBaseContext() const { return CBContext; } 
-   
-   /// Check if the position has any call base context. 
-   bool hasCallBaseContext() const { return CBContext != nullptr; } 
-   
-   /// Special DenseMap key values. 
-   /// 
-   ///{ 
-   static const IRPosition EmptyKey; 
-   static const IRPosition TombstoneKey; 
-   ///} 
-   
-   /// Conversion into a void * to allow reuse of pointer hashing. 
-   operator void *() const { return Enc.getOpaqueValue(); } 
-   
- private: 
-   /// Private constructor for special values only! 
-   explicit IRPosition(void *Ptr, const CallBaseContext *CBContext = nullptr) 
-       : CBContext(CBContext) { 
-     Enc.setFromOpaqueValue(Ptr); 
-   } 
-   
-   /// IRPosition anchored at \p AnchorVal with kind/argument numbet \p PK. 
-   explicit IRPosition(Value &AnchorVal, Kind PK, 
-                       const CallBaseContext *CBContext = nullptr) 
-       : CBContext(CBContext) { 
-     switch (PK) { 
-     case IRPosition::IRP_INVALID: 
-       llvm_unreachable("Cannot create invalid IRP with an anchor value!"); 
-       break; 
-     case IRPosition::IRP_FLOAT: 
-       // Special case for floating functions. 
-       if (isa<Function>(AnchorVal) || isa<CallBase>(AnchorVal)) 
-         Enc = {&AnchorVal, ENC_FLOATING_FUNCTION}; 
-       else 
-         Enc = {&AnchorVal, ENC_VALUE}; 
-       break; 
-     case IRPosition::IRP_FUNCTION: 
-     case IRPosition::IRP_CALL_SITE: 
-       Enc = {&AnchorVal, ENC_VALUE}; 
-       break; 
-     case IRPosition::IRP_RETURNED: 
-     case IRPosition::IRP_CALL_SITE_RETURNED: 
-       Enc = {&AnchorVal, ENC_RETURNED_VALUE}; 
-       break; 
-     case IRPosition::IRP_ARGUMENT: 
-       Enc = {&AnchorVal, ENC_VALUE}; 
-       break; 
-     case IRPosition::IRP_CALL_SITE_ARGUMENT: 
-       llvm_unreachable( 
-           "Cannot create call site argument IRP with an anchor value!"); 
-       break; 
-     } 
-     verify(); 
-   } 
-   
-   /// Return the callee argument number of the associated value if it is an 
-   /// argument or call site argument. See also `getCalleeArgNo` and 
-   /// `getCallSiteArgNo`. 
-   int getArgNo(bool CallbackCalleeArgIfApplicable) const { 
-     if (CallbackCalleeArgIfApplicable) 
-       if (Argument *Arg = getAssociatedArgument()) 
-         return Arg->getArgNo(); 
-     switch (getPositionKind()) { 
-     case IRPosition::IRP_ARGUMENT: 
-       return cast<Argument>(getAsValuePtr())->getArgNo(); 
-     case IRPosition::IRP_CALL_SITE_ARGUMENT: { 
-       Use &U = *getAsUsePtr(); 
-       return cast<CallBase>(U.getUser())->getArgOperandNo(&U); 
-     } 
-     default: 
-       return -1; 
-     } 
-   } 
-   
-   /// IRPosition for the use \p U. The position kind \p PK needs to be 
-   /// IRP_CALL_SITE_ARGUMENT, the anchor value is the user, the associated value 
-   /// the used value. 
-   explicit IRPosition(Use &U, Kind PK) { 
-     assert(PK == IRP_CALL_SITE_ARGUMENT && 
-            "Use constructor is for call site arguments only!"); 
-     Enc = {&U, ENC_CALL_SITE_ARGUMENT_USE}; 
-     verify(); 
-   } 
-   
-   /// Verify internal invariants. 
-   void verify(); 
-   
-   /// Return the attributes of kind \p AK existing in the IR as attribute. 
-   bool getAttrsFromIRAttr(Attribute::AttrKind AK, 
-                           SmallVectorImpl<Attribute> &Attrs) const; 
-   
-   /// Return the attributes of kind \p AK existing in the IR as operand bundles 
-   /// of an llvm.assume. 
-   bool getAttrsFromAssumes(Attribute::AttrKind AK, 
-                            SmallVectorImpl<Attribute> &Attrs, 
-                            Attributor &A) const; 
-   
-   /// Return the underlying pointer as Value *, valid for all positions but 
-   /// IRP_CALL_SITE_ARGUMENT. 
-   Value *getAsValuePtr() const { 
-     assert(getEncodingBits() != ENC_CALL_SITE_ARGUMENT_USE && 
-            "Not a value pointer!"); 
-     return reinterpret_cast<Value *>(Enc.getPointer()); 
-   } 
-   
-   /// Return the underlying pointer as Use *, valid only for 
-   /// IRP_CALL_SITE_ARGUMENT positions. 
-   Use *getAsUsePtr() const { 
-     assert(getEncodingBits() == ENC_CALL_SITE_ARGUMENT_USE && 
-            "Not a value pointer!"); 
-     return reinterpret_cast<Use *>(Enc.getPointer()); 
-   } 
-   
-   /// Return true if \p EncodingBits describe a returned or call site returned 
-   /// position. 
-   static bool isReturnPosition(char EncodingBits) { 
-     return EncodingBits == ENC_RETURNED_VALUE; 
-   } 
-   
-   /// Return true if the encoding bits describe a returned or call site returned 
-   /// position. 
-   bool isReturnPosition() const { return isReturnPosition(getEncodingBits()); } 
-   
-   /// The encoding of the IRPosition is a combination of a pointer and two 
-   /// encoding bits. The values of the encoding bits are defined in the enum 
-   /// below. The pointer is either a Value* (for the first three encoding bit 
-   /// combinations) or Use* (for ENC_CALL_SITE_ARGUMENT_USE). 
-   /// 
-   ///{ 
-   enum { 
-     ENC_VALUE = 0b00, 
-     ENC_RETURNED_VALUE = 0b01, 
-     ENC_FLOATING_FUNCTION = 0b10, 
-     ENC_CALL_SITE_ARGUMENT_USE = 0b11, 
-   }; 
-   
-   // Reserve the maximal amount of bits so there is no need to mask out the 
-   // remaining ones. We will not encode anything else in the pointer anyway. 
-   static constexpr int NumEncodingBits = 
-       PointerLikeTypeTraits<void *>::NumLowBitsAvailable; 
-   static_assert(NumEncodingBits >= 2, "At least two bits are required!"); 
-   
-   /// The pointer with the encoding bits. 
-   PointerIntPair<void *, NumEncodingBits, char> Enc; 
-   ///} 
-   
-   /// Call base context. Used for callsite specific analysis. 
-   const CallBaseContext *CBContext = nullptr; 
-   
-   /// Return the encoding bits. 
-   char getEncodingBits() const { return Enc.getInt(); } 
- }; 
-   
- /// Helper that allows IRPosition as a key in a DenseMap. 
- template <> struct DenseMapInfo<IRPosition> { 
-   static inline IRPosition getEmptyKey() { return IRPosition::EmptyKey; } 
-   static inline IRPosition getTombstoneKey() { 
-     return IRPosition::TombstoneKey; 
-   } 
-   static unsigned getHashValue(const IRPosition &IRP) { 
-     return (DenseMapInfo<void *>::getHashValue(IRP) << 4) ^ 
-            (DenseMapInfo<Value *>::getHashValue(IRP.getCallBaseContext())); 
-   } 
-   
-   static bool isEqual(const IRPosition &a, const IRPosition &b) { 
-     return a == b; 
-   } 
- }; 
-   
- /// A visitor class for IR positions. 
- /// 
- /// Given a position P, the SubsumingPositionIterator allows to visit "subsuming 
- /// positions" wrt. attributes/information. Thus, if a piece of information 
- /// holds for a subsuming position, it also holds for the position P. 
- /// 
- /// The subsuming positions always include the initial position and then, 
- /// depending on the position kind, additionally the following ones: 
- /// - for IRP_RETURNED: 
- ///   - the function (IRP_FUNCTION) 
- /// - for IRP_ARGUMENT: 
- ///   - the function (IRP_FUNCTION) 
- /// - for IRP_CALL_SITE: 
- ///   - the callee (IRP_FUNCTION), if known 
- /// - for IRP_CALL_SITE_RETURNED: 
- ///   - the callee (IRP_RETURNED), if known 
- ///   - the call site (IRP_FUNCTION) 
- ///   - the callee (IRP_FUNCTION), if known 
- /// - for IRP_CALL_SITE_ARGUMENT: 
- ///   - the argument of the callee (IRP_ARGUMENT), if known 
- ///   - the callee (IRP_FUNCTION), if known 
- ///   - the position the call site argument is associated with if it is not 
- ///     anchored to the call site, e.g., if it is an argument then the argument 
- ///     (IRP_ARGUMENT) 
- class SubsumingPositionIterator { 
-   SmallVector<IRPosition, 4> IRPositions; 
-   using iterator = decltype(IRPositions)::iterator; 
-   
- public: 
-   SubsumingPositionIterator(const IRPosition &IRP); 
-   iterator begin() { return IRPositions.begin(); } 
-   iterator end() { return IRPositions.end(); } 
- }; 
-   
- /// Wrapper for FunctionAnalysisManager. 
- struct AnalysisGetter { 
-   // The client may be running the old pass manager, in which case, we need to 
-   // map the requested Analysis to its equivalent wrapper in the old pass 
-   // manager. The scheme implemented here does not require every Analysis to be 
-   // updated. Only those new analyses that the client cares about in the old 
-   // pass manager need to expose a LegacyWrapper type, and that wrapper should 
-   // support a getResult() method that matches the new Analysis. 
-   // 
-   // We need SFINAE to check for the LegacyWrapper, but function templates don't 
-   // allow partial specialization, which is needed in this case. So instead, we 
-   // use a constexpr bool to perform the SFINAE, and then use this information 
-   // inside the function template. 
-   template <typename, typename = void> static constexpr bool HasLegacyWrapper = false; 
-   
-   template <typename Analysis> 
-   typename Analysis::Result *getAnalysis(const Function &F) { 
-     if (FAM) 
-       return &FAM->getResult<Analysis>(const_cast<Function &>(F)); 
-     if constexpr (HasLegacyWrapper<Analysis>) 
-       if (LegacyPass) 
-         return &LegacyPass 
-                     ->getAnalysis<typename Analysis::LegacyWrapper>( 
-                         const_cast<Function &>(F)) 
-                     .getResult(); 
-     return nullptr; 
-   } 
-   
-   AnalysisGetter(FunctionAnalysisManager &FAM) : FAM(&FAM) {} 
-   AnalysisGetter(Pass *P) : LegacyPass(P) {} 
-   AnalysisGetter() = default; 
-   
- private: 
-   FunctionAnalysisManager *FAM = nullptr; 
-   Pass *LegacyPass = nullptr; 
- }; 
-   
- template <typename Analysis> 
- constexpr bool AnalysisGetter::HasLegacyWrapper< 
-       Analysis, std::void_t<typename Analysis::LegacyWrapper>> = true; 
-   
- /// Data structure to hold cached (LLVM-IR) information. 
- /// 
- /// All attributes are given an InformationCache object at creation time to 
- /// avoid inspection of the IR by all of them individually. This default 
- /// InformationCache will hold information required by 'default' attributes, 
- /// thus the ones deduced when Attributor::identifyDefaultAbstractAttributes(..) 
- /// is called. 
- /// 
- /// If custom abstract attributes, registered manually through 
- /// Attributor::registerAA(...), need more information, especially if it is not 
- /// reusable, it is advised to inherit from the InformationCache and cast the 
- /// instance down in the abstract attributes. 
- struct InformationCache { 
-   InformationCache(const Module &M, AnalysisGetter &AG, 
-                    BumpPtrAllocator &Allocator, SetVector<Function *> *CGSCC) 
-       : DL(M.getDataLayout()), Allocator(Allocator), 
-         Explorer( 
-             /* ExploreInterBlock */ true, /* ExploreCFGForward */ true, 
-             /* ExploreCFGBackward */ true, 
-             /* LIGetter */ 
-             [&](const Function &F) { return AG.getAnalysis<LoopAnalysis>(F); }, 
-             /* DTGetter */ 
-             [&](const Function &F) { 
-               return AG.getAnalysis<DominatorTreeAnalysis>(F); 
-             }, 
-             /* PDTGetter */ 
-             [&](const Function &F) { 
-               return AG.getAnalysis<PostDominatorTreeAnalysis>(F); 
-             }), 
-         AG(AG), TargetTriple(M.getTargetTriple()) { 
-     if (CGSCC) 
-       initializeModuleSlice(*CGSCC); 
-   } 
-   
-   ~InformationCache() { 
-     // The FunctionInfo objects are allocated via a BumpPtrAllocator, we call 
-     // the destructor manually. 
-     for (auto &It : FuncInfoMap) 
-       It.getSecond()->~FunctionInfo(); 
-     // Same is true for the instruction exclusions sets. 
-     using AA::InstExclusionSetTy; 
-     for (auto *BES : BESets) 
-       BES->~InstExclusionSetTy(); 
-   } 
-   
-   /// Apply \p CB to all uses of \p F. If \p LookThroughConstantExprUses is 
-   /// true, constant expression users are not given to \p CB but their uses are 
-   /// traversed transitively. 
-   template <typename CBTy> 
-   static void foreachUse(Function &F, CBTy CB, 
-                          bool LookThroughConstantExprUses = true) { 
-     SmallVector<Use *, 8> Worklist(make_pointer_range(F.uses())); 
-   
-     for (unsigned Idx = 0; Idx < Worklist.size(); ++Idx) { 
-       Use &U = *Worklist[Idx]; 
-   
-       // Allow use in constant bitcasts and simply look through them. 
-       if (LookThroughConstantExprUses && isa<ConstantExpr>(U.getUser())) { 
-         for (Use &CEU : cast<ConstantExpr>(U.getUser())->uses()) 
-           Worklist.push_back(&CEU); 
-         continue; 
-       } 
-   
-       CB(U); 
-     } 
-   } 
-   
-   /// Initialize the ModuleSlice member based on \p SCC. ModuleSlices contains 
-   /// (a subset of) all functions that we can look at during this SCC traversal. 
-   /// This includes functions (transitively) called from the SCC and the 
-   /// (transitive) callers of SCC functions. We also can look at a function if 
-   /// there is a "reference edge", i.a., if the function somehow uses (!=calls) 
-   /// a function in the SCC or a caller of a function in the SCC. 
-   void initializeModuleSlice(SetVector<Function *> &SCC) { 
-     ModuleSlice.insert(SCC.begin(), SCC.end()); 
-   
-     SmallPtrSet<Function *, 16> Seen; 
-     SmallVector<Function *, 16> Worklist(SCC.begin(), SCC.end()); 
-     while (!Worklist.empty()) { 
-       Function *F = Worklist.pop_back_val(); 
-       ModuleSlice.insert(F); 
-   
-       for (Instruction &I : instructions(*F)) 
-         if (auto *CB = dyn_cast<CallBase>(&I)) 
-           if (Function *Callee = CB->getCalledFunction()) 
-             if (Seen.insert(Callee).second) 
-               Worklist.push_back(Callee); 
-     } 
-   
-     Seen.clear(); 
-     Worklist.append(SCC.begin(), SCC.end()); 
-     while (!Worklist.empty()) { 
-       Function *F = Worklist.pop_back_val(); 
-       ModuleSlice.insert(F); 
-   
-       // Traverse all transitive uses. 
-       foreachUse(*F, [&](Use &U) { 
-         if (auto *UsrI = dyn_cast<Instruction>(U.getUser())) 
-           if (Seen.insert(UsrI->getFunction()).second) 
-             Worklist.push_back(UsrI->getFunction()); 
-       }); 
-     } 
-   } 
-   
-   /// The slice of the module we are allowed to look at. 
-   SmallPtrSet<Function *, 8> ModuleSlice; 
-   
-   /// A vector type to hold instructions. 
-   using InstructionVectorTy = SmallVector<Instruction *, 8>; 
-   
-   /// A map type from opcodes to instructions with this opcode. 
-   using OpcodeInstMapTy = DenseMap<unsigned, InstructionVectorTy *>; 
-   
-   /// Return the map that relates "interesting" opcodes with all instructions 
-   /// with that opcode in \p F. 
-   OpcodeInstMapTy &getOpcodeInstMapForFunction(const Function &F) { 
-     return getFunctionInfo(F).OpcodeInstMap; 
-   } 
-   
-   /// Return the instructions in \p F that may read or write memory. 
-   InstructionVectorTy &getReadOrWriteInstsForFunction(const Function &F) { 
-     return getFunctionInfo(F).RWInsts; 
-   } 
-   
-   /// Return MustBeExecutedContextExplorer 
-   MustBeExecutedContextExplorer &getMustBeExecutedContextExplorer() { 
-     return Explorer; 
-   } 
-   
-   /// Return TargetLibraryInfo for function \p F. 
-   TargetLibraryInfo *getTargetLibraryInfoForFunction(const Function &F) { 
-     return AG.getAnalysis<TargetLibraryAnalysis>(F); 
-   } 
-   
-   /// Return AliasAnalysis Result for function \p F. 
-   AAResults *getAAResultsForFunction(const Function &F); 
-   
-   /// Return true if \p Arg is involved in a must-tail call, thus the argument 
-   /// of the caller or callee. 
-   bool isInvolvedInMustTailCall(const Argument &Arg) { 
-     FunctionInfo &FI = getFunctionInfo(*Arg.getParent()); 
-     return FI.CalledViaMustTail || FI.ContainsMustTailCall; 
-   } 
-   
-   bool isOnlyUsedByAssume(const Instruction &I) const { 
-     return AssumeOnlyValues.contains(&I); 
-   } 
-   
-   /// Return the analysis result from a pass \p AP for function \p F. 
-   template <typename AP> 
-   typename AP::Result *getAnalysisResultForFunction(const Function &F) { 
-     return AG.getAnalysis<AP>(F); 
-   } 
-   
-   /// Return datalayout used in the module. 
-   const DataLayout &getDL() { return DL; } 
-   
-   /// Return the map conaining all the knowledge we have from `llvm.assume`s. 
-   const RetainedKnowledgeMap &getKnowledgeMap() const { return KnowledgeMap; } 
-   
-   /// Given \p BES, return a uniqued version. \p BES is destroyed in the 
-   /// process. 
-   const AA::InstExclusionSetTy * 
-   getOrCreateUniqueBlockExecutionSet(const AA::InstExclusionSetTy *BES) { 
-     auto It = BESets.find(BES); 
-     if (It != BESets.end()) 
-       return *It; 
-     auto *UniqueBES = new (Allocator) AA::InstExclusionSetTy(*BES); 
-     BESets.insert(UniqueBES); 
-     return UniqueBES; 
-   } 
-   
-   /// Check whether \p F is part of module slice. 
-   bool isInModuleSlice(const Function &F) { 
-     return ModuleSlice.empty() || ModuleSlice.count(const_cast<Function *>(&F)); 
-   } 
-   
-   /// Return true if the stack (llvm::Alloca) can be accessed by other threads. 
-   bool stackIsAccessibleByOtherThreads() { return !targetIsGPU(); } 
-   
-   /// Return true if the target is a GPU. 
-   bool targetIsGPU() { 
-     return TargetTriple.isAMDGPU() || TargetTriple.isNVPTX(); 
-   } 
-   
- private: 
-   struct FunctionInfo { 
-     ~FunctionInfo(); 
-   
-     /// A nested map that remembers all instructions in a function with a 
-     /// certain instruction opcode (Instruction::getOpcode()). 
-     OpcodeInstMapTy OpcodeInstMap; 
-   
-     /// A map from functions to their instructions that may read or write 
-     /// memory. 
-     InstructionVectorTy RWInsts; 
-   
-     /// Function is called by a `musttail` call. 
-     bool CalledViaMustTail; 
-   
-     /// Function contains a `musttail` call. 
-     bool ContainsMustTailCall; 
-   }; 
-   
-   /// A map type from functions to informatio about it. 
-   DenseMap<const Function *, FunctionInfo *> FuncInfoMap; 
-   
-   /// Return information about the function \p F, potentially by creating it. 
-   FunctionInfo &getFunctionInfo(const Function &F) { 
-     FunctionInfo *&FI = FuncInfoMap[&F]; 
-     if (!FI) { 
-       FI = new (Allocator) FunctionInfo(); 
-       initializeInformationCache(F, *FI); 
-     } 
-     return *FI; 
-   } 
-   
-   /// Initialize the function information cache \p FI for the function \p F. 
-   /// 
-   /// This method needs to be called for all function that might be looked at 
-   /// through the information cache interface *prior* to looking at them. 
-   void initializeInformationCache(const Function &F, FunctionInfo &FI); 
-   
-   /// The datalayout used in the module. 
-   const DataLayout &DL; 
-   
-   /// The allocator used to allocate memory, e.g. for `FunctionInfo`s. 
-   BumpPtrAllocator &Allocator; 
-   
-   /// MustBeExecutedContextExplorer 
-   MustBeExecutedContextExplorer Explorer; 
-   
-   /// A map with knowledge retained in `llvm.assume` instructions. 
-   RetainedKnowledgeMap KnowledgeMap; 
-   
-   /// A container for all instructions that are only used by `llvm.assume`. 
-   SetVector<const Instruction *> AssumeOnlyValues; 
-   
-   /// Cache for block sets to allow reuse. 
-   DenseSet<AA::InstExclusionSetTy *> BESets; 
-   
-   /// Getters for analysis. 
-   AnalysisGetter &AG; 
-   
-   /// Set of inlineable functions 
-   SmallPtrSet<const Function *, 8> InlineableFunctions; 
-   
-   /// The triple describing the target machine. 
-   Triple TargetTriple; 
-   
-   /// Give the Attributor access to the members so 
-   /// Attributor::identifyDefaultAbstractAttributes(...) can initialize them. 
-   friend struct Attributor; 
- }; 
-   
- /// Configuration for the Attributor. 
- struct AttributorConfig { 
-   
-   AttributorConfig(CallGraphUpdater &CGUpdater) : CGUpdater(CGUpdater) {} 
-   
-   /// Is the user of the Attributor a module pass or not. This determines what 
-   /// IR we can look at and modify. If it is a module pass we might deduce facts 
-   /// outside the initial function set and modify functions outside that set, 
-   /// but only as part of the optimization of the functions in the initial 
-   /// function set. For CGSCC passes we can look at the IR of the module slice 
-   /// but never run any deduction, or perform any modification, outside the 
-   /// initial function set (which we assume is the SCC). 
-   bool IsModulePass = true; 
-   
-   /// Flag to determine if we can delete functions or keep dead ones around. 
-   bool DeleteFns = true; 
-   
-   /// Flag to determine if we rewrite function signatures. 
-   bool RewriteSignatures = true; 
-   
-   /// Flag to determine if we want to initialize all default AAs for an internal 
-   /// function marked live. See also: InitializationCallback> 
-   bool DefaultInitializeLiveInternals = true; 
-   
-   /// Callback function to be invoked on internal functions marked live. 
-   std::function<void(Attributor &A, const Function &F)> InitializationCallback = 
-       nullptr; 
-   
-   /// Helper to update an underlying call graph and to delete functions. 
-   CallGraphUpdater &CGUpdater; 
-   
-   /// If not null, a set limiting the attribute opportunities. 
-   DenseSet<const char *> *Allowed = nullptr; 
-   
-   /// Maximum number of iterations to run until fixpoint. 
-   std::optional<unsigned> MaxFixpointIterations; 
-   
-   /// A callback function that returns an ORE object from a Function pointer. 
-   ///{ 
-   using OptimizationRemarkGetter = 
-       function_ref<OptimizationRemarkEmitter &(Function *)>; 
-   OptimizationRemarkGetter OREGetter = nullptr; 
-   ///} 
-   
-   /// The name of the pass running the attributor, used to emit remarks. 
-   const char *PassName = nullptr; 
- }; 
-   
- /// The fixpoint analysis framework that orchestrates the attribute deduction. 
- /// 
- /// The Attributor provides a general abstract analysis framework (guided 
- /// fixpoint iteration) as well as helper functions for the deduction of 
- /// (LLVM-IR) attributes. However, also other code properties can be deduced, 
- /// propagated, and ultimately manifested through the Attributor framework. This 
- /// is particularly useful if these properties interact with attributes and a 
- /// co-scheduled deduction allows to improve the solution. Even if not, thus if 
- /// attributes/properties are completely isolated, they should use the 
- /// Attributor framework to reduce the number of fixpoint iteration frameworks 
- /// in the code base. Note that the Attributor design makes sure that isolated 
- /// attributes are not impacted, in any way, by others derived at the same time 
- /// if there is no cross-reasoning performed. 
- /// 
- /// The public facing interface of the Attributor is kept simple and basically 
- /// allows abstract attributes to one thing, query abstract attributes 
- /// in-flight. There are two reasons to do this: 
- ///    a) The optimistic state of one abstract attribute can justify an 
- ///       optimistic state of another, allowing to framework to end up with an 
- ///       optimistic (=best possible) fixpoint instead of one based solely on 
- ///       information in the IR. 
- ///    b) This avoids reimplementing various kinds of lookups, e.g., to check 
- ///       for existing IR attributes, in favor of a single lookups interface 
- ///       provided by an abstract attribute subclass. 
- /// 
- /// NOTE: The mechanics of adding a new "concrete" abstract attribute are 
- ///       described in the file comment. 
- struct Attributor { 
-   
-   /// Constructor 
-   /// 
-   /// \param Functions The set of functions we are deriving attributes for. 
-   /// \param InfoCache Cache to hold various information accessible for 
-   ///                  the abstract attributes. 
-   /// \param Configuration The Attributor configuration which determines what 
-   ///                      generic features to use. 
-   Attributor(SetVector<Function *> &Functions, InformationCache &InfoCache, 
-              AttributorConfig Configuration) 
-       : Allocator(InfoCache.Allocator), Functions(Functions), 
-         InfoCache(InfoCache), Configuration(Configuration) {} 
-   
-   ~Attributor(); 
-   
-   /// Run the analyses until a fixpoint is reached or enforced (timeout). 
-   /// 
-   /// The attributes registered with this Attributor can be used after as long 
-   /// as the Attributor is not destroyed (it owns the attributes now). 
-   /// 
-   /// \Returns CHANGED if the IR was changed, otherwise UNCHANGED. 
-   ChangeStatus run(); 
-   
-   /// Lookup an abstract attribute of type \p AAType at position \p IRP. While 
-   /// no abstract attribute is found equivalent positions are checked, see 
-   /// SubsumingPositionIterator. Thus, the returned abstract attribute 
-   /// might be anchored at a different position, e.g., the callee if \p IRP is a 
-   /// call base. 
-   /// 
-   /// This method is the only (supported) way an abstract attribute can retrieve 
-   /// information from another abstract attribute. As an example, take an 
-   /// abstract attribute that determines the memory access behavior for a 
-   /// argument (readnone, readonly, ...). It should use `getAAFor` to get the 
-   /// most optimistic information for other abstract attributes in-flight, e.g. 
-   /// the one reasoning about the "captured" state for the argument or the one 
-   /// reasoning on the memory access behavior of the function as a whole. 
-   /// 
-   /// If the DepClass enum is set to `DepClassTy::None` the dependence from 
-   /// \p QueryingAA to the return abstract attribute is not automatically 
-   /// recorded. This should only be used if the caller will record the 
-   /// dependence explicitly if necessary, thus if it the returned abstract 
-   /// attribute is used for reasoning. To record the dependences explicitly use 
-   /// the `Attributor::recordDependence` method. 
-   template <typename AAType> 
-   const AAType &getAAFor(const AbstractAttribute &QueryingAA, 
-                          const IRPosition &IRP, DepClassTy DepClass) { 
-     return getOrCreateAAFor<AAType>(IRP, &QueryingAA, DepClass, 
-                                     /* ForceUpdate */ false); 
-   } 
-   
-   /// Similar to getAAFor but the return abstract attribute will be updated (via 
-   /// `AbstractAttribute::update`) even if it is found in the cache. This is 
-   /// especially useful for AAIsDead as changes in liveness can make updates 
-   /// possible/useful that were not happening before as the abstract attribute 
-   /// was assumed dead. 
-   template <typename AAType> 
-   const AAType &getAndUpdateAAFor(const AbstractAttribute &QueryingAA, 
-                                   const IRPosition &IRP, DepClassTy DepClass) { 
-     return getOrCreateAAFor<AAType>(IRP, &QueryingAA, DepClass, 
-                                     /* ForceUpdate */ true); 
-   } 
-   
-   /// The version of getAAFor that allows to omit a querying abstract 
-   /// attribute. Using this after Attributor started running is restricted to 
-   /// only the Attributor itself. Initial seeding of AAs can be done via this 
-   /// function. 
-   /// NOTE: ForceUpdate is ignored in any stage other than the update stage. 
-   template <typename AAType> 
-   const AAType &getOrCreateAAFor(IRPosition IRP, 
-                                  const AbstractAttribute *QueryingAA, 
-                                  DepClassTy DepClass, bool ForceUpdate = false, 
-                                  bool UpdateAfterInit = true) { 
-     if (!shouldPropagateCallBaseContext(IRP)) 
-       IRP = IRP.stripCallBaseContext(); 
-   
-     if (AAType *AAPtr = lookupAAFor<AAType>(IRP, QueryingAA, DepClass, 
-                                             /* AllowInvalidState */ true)) { 
-       if (ForceUpdate && Phase == AttributorPhase::UPDATE) 
-         updateAA(*AAPtr); 
-       return *AAPtr; 
-     } 
-   
-     // No matching attribute found, create one. 
-     // Use the static create method. 
-     auto &AA = AAType::createForPosition(IRP, *this); 
-   
-     // Always register a new attribute to make sure we clean up the allocated 
-     // memory properly. 
-     registerAA(AA); 
-   
-     // If we are currenty seeding attributes, enforce seeding rules. 
-     if (Phase == AttributorPhase::SEEDING && !shouldSeedAttribute(AA)) { 
-       AA.getState().indicatePessimisticFixpoint(); 
-       return AA; 
-     } 
-   
-     // For now we ignore naked and optnone functions. 
-     bool Invalidate = 
-         Configuration.Allowed && !Configuration.Allowed->count(&AAType::ID); 
-     const Function *AnchorFn = IRP.getAnchorScope(); 
-     if (AnchorFn) { 
-       Invalidate |= 
-           AnchorFn->hasFnAttribute(Attribute::Naked) || 
-           AnchorFn->hasFnAttribute(Attribute::OptimizeNone) || 
-           (!isModulePass() && !getInfoCache().isInModuleSlice(*AnchorFn)); 
-     } 
-   
-     // Avoid too many nested initializations to prevent a stack overflow. 
-     Invalidate |= InitializationChainLength > MaxInitializationChainLength; 
-   
-     // Bootstrap the new attribute with an initial update to propagate 
-     // information, e.g., function -> call site. If it is not on a given 
-     // Allowed we will not perform updates at all. 
-     if (Invalidate) { 
-       AA.getState().indicatePessimisticFixpoint(); 
-       return AA; 
-     } 
-   
-     { 
-       TimeTraceScope TimeScope(AA.getName() + "::initialize"); 
-       ++InitializationChainLength; 
-       AA.initialize(*this); 
-       --InitializationChainLength; 
-     } 
-   
-     // We update only AAs associated with functions in the Functions set or 
-     // call sites of them. 
-     if ((AnchorFn && !isRunOn(const_cast<Function *>(AnchorFn))) && 
-         !isRunOn(IRP.getAssociatedFunction())) { 
-       AA.getState().indicatePessimisticFixpoint(); 
-       return AA; 
-     } 
-   
-     // If this is queried in the manifest stage, we force the AA to indicate 
-     // pessimistic fixpoint immediately. 
-     if (Phase == AttributorPhase::MANIFEST || 
-         Phase == AttributorPhase::CLEANUP) { 
-       AA.getState().indicatePessimisticFixpoint(); 
-       return AA; 
-     } 
-   
-     // Allow seeded attributes to declare dependencies. 
-     // Remember the seeding state. 
-     if (UpdateAfterInit) { 
-       AttributorPhase OldPhase = Phase; 
-       Phase = AttributorPhase::UPDATE; 
-   
-       updateAA(AA); 
-   
-       Phase = OldPhase; 
-     } 
-   
-     if (QueryingAA && AA.getState().isValidState()) 
-       recordDependence(AA, const_cast<AbstractAttribute &>(*QueryingAA), 
-                        DepClass); 
-     return AA; 
-   } 
-   template <typename AAType> 
-   const AAType &getOrCreateAAFor(const IRPosition &IRP) { 
-     return getOrCreateAAFor<AAType>(IRP, /* QueryingAA */ nullptr, 
-                                     DepClassTy::NONE); 
-   } 
-   
-   /// Return the attribute of \p AAType for \p IRP if existing and valid. This 
-   /// also allows non-AA users lookup. 
-   template <typename AAType> 
-   AAType *lookupAAFor(const IRPosition &IRP, 
-                       const AbstractAttribute *QueryingAA = nullptr, 
-                       DepClassTy DepClass = DepClassTy::OPTIONAL, 
-                       bool AllowInvalidState = false) { 
-     static_assert(std::is_base_of<AbstractAttribute, AAType>::value, 
-                   "Cannot query an attribute with a type not derived from " 
-                   "'AbstractAttribute'!"); 
-     // Lookup the abstract attribute of type AAType. If found, return it after 
-     // registering a dependence of QueryingAA on the one returned attribute. 
-     AbstractAttribute *AAPtr = AAMap.lookup({&AAType::ID, IRP}); 
-     if (!AAPtr) 
-       return nullptr; 
-   
-     AAType *AA = static_cast<AAType *>(AAPtr); 
-   
-     // Do not register a dependence on an attribute with an invalid state. 
-     if (DepClass != DepClassTy::NONE && QueryingAA && 
-         AA->getState().isValidState()) 
-       recordDependence(*AA, const_cast<AbstractAttribute &>(*QueryingAA), 
-                        DepClass); 
-   
-     // Return nullptr if this attribute has an invalid state. 
-     if (!AllowInvalidState && !AA->getState().isValidState()) 
-       return nullptr; 
-     return AA; 
-   } 
-   
-   /// Allows a query AA to request an update if a new query was received. 
-   void registerForUpdate(AbstractAttribute &AA); 
-   
-   /// Explicitly record a dependence from \p FromAA to \p ToAA, that is if 
-   /// \p FromAA changes \p ToAA should be updated as well. 
-   /// 
-   /// This method should be used in conjunction with the `getAAFor` method and 
-   /// with the DepClass enum passed to the method set to None. This can 
-   /// be beneficial to avoid false dependences but it requires the users of 
-   /// `getAAFor` to explicitly record true dependences through this method. 
-   /// The \p DepClass flag indicates if the dependence is striclty necessary. 
-   /// That means for required dependences, if \p FromAA changes to an invalid 
-   /// state, \p ToAA can be moved to a pessimistic fixpoint because it required 
-   /// information from \p FromAA but none are available anymore. 
-   void recordDependence(const AbstractAttribute &FromAA, 
-                         const AbstractAttribute &ToAA, DepClassTy DepClass); 
-   
-   /// Introduce a new abstract attribute into the fixpoint analysis. 
-   /// 
-   /// Note that ownership of the attribute is given to the Attributor. It will 
-   /// invoke delete for the Attributor on destruction of the Attributor. 
-   /// 
-   /// Attributes are identified by their IR position (AAType::getIRPosition()) 
-   /// and the address of their static member (see AAType::ID). 
-   template <typename AAType> AAType ®isterAA(AAType &AA) { 
-     static_assert(std::is_base_of<AbstractAttribute, AAType>::value, 
-                   "Cannot register an attribute with a type not derived from " 
-                   "'AbstractAttribute'!"); 
-     // Put the attribute in the lookup map structure and the container we use to 
-     // keep track of all attributes. 
-     const IRPosition &IRP = AA.getIRPosition(); 
-     AbstractAttribute *&AAPtr = AAMap[{&AAType::ID, IRP}]; 
-   
-     assert(!AAPtr && "Attribute already in map!"); 
-     AAPtr = &AA; 
-   
-     // Register AA with the synthetic root only before the manifest stage. 
-     if (Phase == AttributorPhase::SEEDING || Phase == AttributorPhase::UPDATE) 
-       DG.SyntheticRoot.Deps.push_back( 
-           AADepGraphNode::DepTy(&AA, unsigned(DepClassTy::REQUIRED))); 
-   
-     return AA; 
-   } 
-   
-   /// Return the internal information cache. 
-   InformationCache &getInfoCache() { return InfoCache; } 
-   
-   /// Return true if this is a module pass, false otherwise. 
-   bool isModulePass() const { return Configuration.IsModulePass; } 
-   
-   /// Return true if we derive attributes for \p Fn 
-   bool isRunOn(Function &Fn) const { return isRunOn(&Fn); } 
-   bool isRunOn(Function *Fn) const { 
-     return Functions.empty() || Functions.count(Fn); 
-   } 
-   
-   /// Determine opportunities to derive 'default' attributes in \p F and create 
-   /// abstract attribute objects for them. 
-   /// 
-   /// \param F The function that is checked for attribute opportunities. 
-   /// 
-   /// Note that abstract attribute instances are generally created even if the 
-   /// IR already contains the information they would deduce. The most important 
-   /// reason for this is the single interface, the one of the abstract attribute 
-   /// instance, which can be queried without the need to look at the IR in 
-   /// various places. 
-   void identifyDefaultAbstractAttributes(Function &F); 
-   
-   /// Determine whether the function \p F is IPO amendable 
-   /// 
-   /// If a function is exactly defined or it has alwaysinline attribute 
-   /// and is viable to be inlined, we say it is IPO amendable 
-   bool isFunctionIPOAmendable(const Function &F) { 
-     return F.hasExactDefinition() || InfoCache.InlineableFunctions.count(&F); 
-   } 
-   
-   /// Mark the internal function \p F as live. 
-   /// 
-   /// This will trigger the identification and initialization of attributes for 
-   /// \p F. 
-   void markLiveInternalFunction(const Function &F) { 
-     assert(F.hasLocalLinkage() && 
-            "Only local linkage is assumed dead initially."); 
-   
-     if (Configuration.DefaultInitializeLiveInternals) 
-       identifyDefaultAbstractAttributes(const_cast<Function &>(F)); 
-     if (Configuration.InitializationCallback) 
-       Configuration.InitializationCallback(*this, F); 
-   } 
-   
-   /// Helper function to remove callsite. 
-   void removeCallSite(CallInst *CI) { 
-     if (!CI) 
-       return; 
-   
-     Configuration.CGUpdater.removeCallSite(*CI); 
-   } 
-   
-   /// Record that \p U is to be replaces with \p NV after information was 
-   /// manifested. This also triggers deletion of trivially dead istructions. 
-   bool changeUseAfterManifest(Use &U, Value &NV) { 
-     Value *&V = ToBeChangedUses[&U]; 
-     if (V && (V->stripPointerCasts() == NV.stripPointerCasts() || 
-               isa_and_nonnull<UndefValue>(V))) 
-       return false; 
-     assert((!V || V == &NV || isa<UndefValue>(NV)) && 
-            "Use was registered twice for replacement with different values!"); 
-     V = &NV; 
-     return true; 
-   } 
-   
-   /// Helper function to replace all uses associated with \p IRP with \p NV. 
-   /// Return true if there is any change. The flag \p ChangeDroppable indicates 
-   /// if dropppable uses should be changed too. 
-   bool changeAfterManifest(const IRPosition IRP, Value &NV, 
-                            bool ChangeDroppable = true) { 
-     if (IRP.getPositionKind() == IRPosition::IRP_CALL_SITE_ARGUMENT) { 
-       auto *CB = cast<CallBase>(IRP.getCtxI()); 
-       return changeUseAfterManifest( 
-           CB->getArgOperandUse(IRP.getCallSiteArgNo()), NV); 
-     } 
-     Value &V = IRP.getAssociatedValue(); 
-     auto &Entry = ToBeChangedValues[&V]; 
-     Value *CurNV = get<0>(Entry); 
-     if (CurNV && (CurNV->stripPointerCasts() == NV.stripPointerCasts() || 
-                   isa<UndefValue>(CurNV))) 
-       return false; 
-     assert((!CurNV || CurNV == &NV || isa<UndefValue>(NV)) && 
-            "Value replacement was registered twice with different values!"); 
-     Entry = {&NV, ChangeDroppable}; 
-     return true; 
-   } 
-   
-   /// Record that \p I is to be replaced with `unreachable` after information 
-   /// was manifested. 
-   void changeToUnreachableAfterManifest(Instruction *I) { 
-     ToBeChangedToUnreachableInsts.insert(I); 
-   } 
-   
-   /// Record that \p II has at least one dead successor block. This information 
-   /// is used, e.g., to replace \p II with a call, after information was 
-   /// manifested. 
-   void registerInvokeWithDeadSuccessor(InvokeInst &II) { 
-     InvokeWithDeadSuccessor.insert(&II); 
-   } 
-   
-   /// Record that \p I is deleted after information was manifested. This also 
-   /// triggers deletion of trivially dead istructions. 
-   void deleteAfterManifest(Instruction &I) { ToBeDeletedInsts.insert(&I); } 
-   
-   /// Record that \p BB is deleted after information was manifested. This also 
-   /// triggers deletion of trivially dead istructions. 
-   void deleteAfterManifest(BasicBlock &BB) { ToBeDeletedBlocks.insert(&BB); } 
-   
-   // Record that \p BB is added during the manifest of an AA. Added basic blocks 
-   // are preserved in the IR. 
-   void registerManifestAddedBasicBlock(BasicBlock &BB) { 
-     ManifestAddedBlocks.insert(&BB); 
-   } 
-   
-   /// Record that \p F is deleted after information was manifested. 
-   void deleteAfterManifest(Function &F) { 
-     if (Configuration.DeleteFns) 
-       ToBeDeletedFunctions.insert(&F); 
-   } 
-   
-   /// If \p IRP is assumed to be a constant, return it, if it is unclear yet, 
-   /// return std::nullopt, otherwise return `nullptr`. 
-   std::optional<Constant *> getAssumedConstant(const IRPosition &IRP, 
-                                                const AbstractAttribute &AA, 
-                                                bool &UsedAssumedInformation); 
-   std::optional<Constant *> getAssumedConstant(const Value &V, 
-                                                const AbstractAttribute &AA, 
-                                                bool &UsedAssumedInformation) { 
-     return getAssumedConstant(IRPosition::value(V), AA, UsedAssumedInformation); 
-   } 
-   
-   /// If \p V is assumed simplified, return it, if it is unclear yet, 
-   /// return std::nullopt, otherwise return `nullptr`. 
-   std::optional<Value *> getAssumedSimplified(const IRPosition &IRP, 
-                                               const AbstractAttribute &AA, 
-                                               bool &UsedAssumedInformation, 
-                                               AA::ValueScope S) { 
-     return getAssumedSimplified(IRP, &AA, UsedAssumedInformation, S); 
-   } 
-   std::optional<Value *> getAssumedSimplified(const Value &V, 
-                                               const AbstractAttribute &AA, 
-                                               bool &UsedAssumedInformation, 
-                                               AA::ValueScope S) { 
-     return getAssumedSimplified(IRPosition::value(V), AA, 
-                                 UsedAssumedInformation, S); 
-   } 
-   
-   /// If \p V is assumed simplified, return it, if it is unclear yet, 
-   /// return std::nullopt, otherwise return `nullptr`. Same as the public 
-   /// version except that it can be used without recording dependences on any \p 
-   /// AA. 
-   std::optional<Value *> getAssumedSimplified(const IRPosition &V, 
-                                               const AbstractAttribute *AA, 
-                                               bool &UsedAssumedInformation, 
-                                               AA::ValueScope S); 
-   
-   /// Try to simplify \p IRP and in the scope \p S. If successful, true is 
-   /// returned and all potential values \p IRP can take are put into \p Values. 
-   /// If the result in \p Values contains select or PHI instructions it means 
-   /// those could not be simplified to a single value. Recursive calls with 
-   /// these instructions will yield their respective potential values. If false 
-   /// is returned no other information is valid. 
-   bool getAssumedSimplifiedValues(const IRPosition &IRP, 
-                                   const AbstractAttribute *AA, 
-                                   SmallVectorImpl<AA::ValueAndContext> &Values, 
-                                   AA::ValueScope S, 
-                                   bool &UsedAssumedInformation); 
-   
-   /// Register \p CB as a simplification callback. 
-   /// `Attributor::getAssumedSimplified` will use these callbacks before 
-   /// we it will ask `AAValueSimplify`. It is important to ensure this 
-   /// is called before `identifyDefaultAbstractAttributes`, assuming the 
-   /// latter is called at all. 
-   using SimplifictionCallbackTy = std::function<std::optional<Value *>( 
-       const IRPosition &, const AbstractAttribute *, bool &)>; 
-   void registerSimplificationCallback(const IRPosition &IRP, 
-                                       const SimplifictionCallbackTy &CB) { 
-     SimplificationCallbacks[IRP].emplace_back(CB); 
-   } 
-   
-   /// Return true if there is a simplification callback for \p IRP. 
-   bool hasSimplificationCallback(const IRPosition &IRP) { 
-     return SimplificationCallbacks.count(IRP); 
-   } 
-   
-   using VirtualUseCallbackTy = 
-       std::function<bool(Attributor &, const AbstractAttribute *)>; 
-   void registerVirtualUseCallback(const Value &V, 
-                                   const VirtualUseCallbackTy &CB) { 
-     VirtualUseCallbacks[&V].emplace_back(CB); 
-   } 
-   
- private: 
-   /// The vector with all simplification callbacks registered by outside AAs. 
-   DenseMap<IRPosition, SmallVector<SimplifictionCallbackTy, 1>> 
-       SimplificationCallbacks; 
-   
-   DenseMap<const Value *, SmallVector<VirtualUseCallbackTy, 1>> 
-       VirtualUseCallbacks; 
-   
- public: 
-   /// Translate \p V from the callee context into the call site context. 
-   std::optional<Value *> 
-   translateArgumentToCallSiteContent(std::optional<Value *> V, CallBase &CB, 
-                                      const AbstractAttribute &AA, 
-                                      bool &UsedAssumedInformation); 
-   
-   /// Return true if \p AA (or its context instruction) is assumed dead. 
-   /// 
-   /// If \p LivenessAA is not provided it is queried. 
-   bool isAssumedDead(const AbstractAttribute &AA, const AAIsDead *LivenessAA, 
-                      bool &UsedAssumedInformation, 
-                      bool CheckBBLivenessOnly = false, 
-                      DepClassTy DepClass = DepClassTy::OPTIONAL); 
-   
-   /// Return true if \p I is assumed dead. 
-   /// 
-   /// If \p LivenessAA is not provided it is queried. 
-   bool isAssumedDead(const Instruction &I, const AbstractAttribute *QueryingAA, 
-                      const AAIsDead *LivenessAA, bool &UsedAssumedInformation, 
-                      bool CheckBBLivenessOnly = false, 
-                      DepClassTy DepClass = DepClassTy::OPTIONAL, 
-                      bool CheckForDeadStore = false); 
-   
-   /// Return true if \p U is assumed dead. 
-   /// 
-   /// If \p FnLivenessAA is not provided it is queried. 
-   bool isAssumedDead(const Use &U, const AbstractAttribute *QueryingAA, 
-                      const AAIsDead *FnLivenessAA, bool &UsedAssumedInformation, 
-                      bool CheckBBLivenessOnly = false, 
-                      DepClassTy DepClass = DepClassTy::OPTIONAL); 
-   
-   /// Return true if \p IRP is assumed dead. 
-   /// 
-   /// If \p FnLivenessAA is not provided it is queried. 
-   bool isAssumedDead(const IRPosition &IRP, const AbstractAttribute *QueryingAA, 
-                      const AAIsDead *FnLivenessAA, bool &UsedAssumedInformation, 
-                      bool CheckBBLivenessOnly = false, 
-                      DepClassTy DepClass = DepClassTy::OPTIONAL); 
-   
-   /// Return true if \p BB is assumed dead. 
-   /// 
-   /// If \p LivenessAA is not provided it is queried. 
-   bool isAssumedDead(const BasicBlock &BB, const AbstractAttribute *QueryingAA, 
-                      const AAIsDead *FnLivenessAA, 
-                      DepClassTy DepClass = DepClassTy::OPTIONAL); 
-   
-   /// Check \p Pred on all (transitive) uses of \p V. 
-   /// 
-   /// This method will evaluate \p Pred on all (transitive) uses of the 
-   /// associated value and return true if \p Pred holds every time. 
-   /// If uses are skipped in favor of equivalent ones, e.g., if we look through 
-   /// memory, the \p EquivalentUseCB will be used to give the caller an idea 
-   /// what original used was replaced by a new one (or new ones). The visit is 
-   /// cut short if \p EquivalentUseCB returns false and the function will return 
-   /// false as well. 
-   bool checkForAllUses(function_ref<bool(const Use &, bool &)> Pred, 
-                        const AbstractAttribute &QueryingAA, const Value &V, 
-                        bool CheckBBLivenessOnly = false, 
-                        DepClassTy LivenessDepClass = DepClassTy::OPTIONAL, 
-                        bool IgnoreDroppableUses = true, 
-                        function_ref<bool(const Use &OldU, const Use &NewU)> 
-                            EquivalentUseCB = nullptr); 
-   
-   /// Emit a remark generically. 
-   /// 
-   /// This template function can be used to generically emit a remark. The 
-   /// RemarkKind should be one of the following: 
-   ///   - OptimizationRemark to indicate a successful optimization attempt 
-   ///   - OptimizationRemarkMissed to report a failed optimization attempt 
-   ///   - OptimizationRemarkAnalysis to provide additional information about an 
-   ///     optimization attempt 
-   /// 
-   /// The remark is built using a callback function \p RemarkCB that takes a 
-   /// RemarkKind as input and returns a RemarkKind. 
-   template <typename RemarkKind, typename RemarkCallBack> 
-   void emitRemark(Instruction *I, StringRef RemarkName, 
-                   RemarkCallBack &&RemarkCB) const { 
-     if (!Configuration.OREGetter) 
-       return; 
-   
-     Function *F = I->getFunction(); 
-     auto &ORE = Configuration.OREGetter(F); 
-   
-     if (RemarkName.startswith("OMP")) 
-       ORE.emit([&]() { 
-         return RemarkCB(RemarkKind(Configuration.PassName, RemarkName, I)) 
-                << " [" << RemarkName << "]"; 
-       }); 
-     else 
-       ORE.emit([&]() { 
-         return RemarkCB(RemarkKind(Configuration.PassName, RemarkName, I)); 
-       }); 
-   } 
-   
-   /// Emit a remark on a function. 
-   template <typename RemarkKind, typename RemarkCallBack> 
-   void emitRemark(Function *F, StringRef RemarkName, 
-                   RemarkCallBack &&RemarkCB) const { 
-     if (!Configuration.OREGetter) 
-       return; 
-   
-     auto &ORE = Configuration.OREGetter(F); 
-   
-     if (RemarkName.startswith("OMP")) 
-       ORE.emit([&]() { 
-         return RemarkCB(RemarkKind(Configuration.PassName, RemarkName, F)) 
-                << " [" << RemarkName << "]"; 
-       }); 
-     else 
-       ORE.emit([&]() { 
-         return RemarkCB(RemarkKind(Configuration.PassName, RemarkName, F)); 
-       }); 
-   } 
-   
-   /// Helper struct used in the communication between an abstract attribute (AA) 
-   /// that wants to change the signature of a function and the Attributor which 
-   /// applies the changes. The struct is partially initialized with the 
-   /// information from the AA (see the constructor). All other members are 
-   /// provided by the Attributor prior to invoking any callbacks. 
-   struct ArgumentReplacementInfo { 
-     /// Callee repair callback type 
-     /// 
-     /// The function repair callback is invoked once to rewire the replacement 
-     /// arguments in the body of the new function. The argument replacement info 
-     /// is passed, as build from the registerFunctionSignatureRewrite call, as 
-     /// well as the replacement function and an iteratore to the first 
-     /// replacement argument. 
-     using CalleeRepairCBTy = std::function<void( 
-         const ArgumentReplacementInfo &, Function &, Function::arg_iterator)>; 
-   
-     /// Abstract call site (ACS) repair callback type 
-     /// 
-     /// The abstract call site repair callback is invoked once on every abstract 
-     /// call site of the replaced function (\see ReplacedFn). The callback needs 
-     /// to provide the operands for the call to the new replacement function. 
-     /// The number and type of the operands appended to the provided vector 
-     /// (second argument) is defined by the number and types determined through 
-     /// the replacement type vector (\see ReplacementTypes). The first argument 
-     /// is the ArgumentReplacementInfo object registered with the Attributor 
-     /// through the registerFunctionSignatureRewrite call. 
-     using ACSRepairCBTy = 
-         std::function<void(const ArgumentReplacementInfo &, AbstractCallSite, 
-                            SmallVectorImpl<Value *> &)>; 
-   
-     /// Simple getters, see the corresponding members for details. 
-     ///{ 
-   
-     Attributor &getAttributor() const { return A; } 
-     const Function &getReplacedFn() const { return ReplacedFn; } 
-     const Argument &getReplacedArg() const { return ReplacedArg; } 
-     unsigned getNumReplacementArgs() const { return ReplacementTypes.size(); } 
-     const SmallVectorImpl<Type *> &getReplacementTypes() const { 
-       return ReplacementTypes; 
-     } 
-   
-     ///} 
-   
-   private: 
-     /// Constructor that takes the argument to be replaced, the types of 
-     /// the replacement arguments, as well as callbacks to repair the call sites 
-     /// and new function after the replacement happened. 
-     ArgumentReplacementInfo(Attributor &A, Argument &Arg, 
-                             ArrayRef<Type *> ReplacementTypes, 
-                             CalleeRepairCBTy &&CalleeRepairCB, 
-                             ACSRepairCBTy &&ACSRepairCB) 
-         : A(A), ReplacedFn(*Arg.getParent()), ReplacedArg(Arg), 
-           ReplacementTypes(ReplacementTypes.begin(), ReplacementTypes.end()), 
-           CalleeRepairCB(std::move(CalleeRepairCB)), 
-           ACSRepairCB(std::move(ACSRepairCB)) {} 
-   
-     /// Reference to the attributor to allow access from the callbacks. 
-     Attributor &A; 
-   
-     /// The "old" function replaced by ReplacementFn. 
-     const Function &ReplacedFn; 
-   
-     /// The "old" argument replaced by new ones defined via ReplacementTypes. 
-     const Argument &ReplacedArg; 
-   
-     /// The types of the arguments replacing ReplacedArg. 
-     const SmallVector<Type *, 8> ReplacementTypes; 
-   
-     /// Callee repair callback, see CalleeRepairCBTy. 
-     const CalleeRepairCBTy CalleeRepairCB; 
-   
-     /// Abstract call site (ACS) repair callback, see ACSRepairCBTy. 
-     const ACSRepairCBTy ACSRepairCB; 
-   
-     /// Allow access to the private members from the Attributor. 
-     friend struct Attributor; 
-   }; 
-   
-   /// Check if we can rewrite a function signature. 
-   /// 
-   /// The argument \p Arg is replaced with new ones defined by the number, 
-   /// order, and types in \p ReplacementTypes. 
-   /// 
-   /// \returns True, if the replacement can be registered, via 
-   /// registerFunctionSignatureRewrite, false otherwise. 
-   bool isValidFunctionSignatureRewrite(Argument &Arg, 
-                                        ArrayRef<Type *> ReplacementTypes); 
-   
-   /// Register a rewrite for a function signature. 
-   /// 
-   /// The argument \p Arg is replaced with new ones defined by the number, 
-   /// order, and types in \p ReplacementTypes. The rewiring at the call sites is 
-   /// done through \p ACSRepairCB and at the callee site through 
-   /// \p CalleeRepairCB. 
-   /// 
-   /// \returns True, if the replacement was registered, false otherwise. 
-   bool registerFunctionSignatureRewrite( 
-       Argument &Arg, ArrayRef<Type *> ReplacementTypes, 
-       ArgumentReplacementInfo::CalleeRepairCBTy &&CalleeRepairCB, 
-       ArgumentReplacementInfo::ACSRepairCBTy &&ACSRepairCB); 
-   
-   /// Check \p Pred on all function call sites. 
-   /// 
-   /// This method will evaluate \p Pred on call sites and return 
-   /// true if \p Pred holds in every call sites. However, this is only possible 
-   /// all call sites are known, hence the function has internal linkage. 
-   /// If true is returned, \p UsedAssumedInformation is set if assumed 
-   /// information was used to skip or simplify potential call sites. 
-   bool checkForAllCallSites(function_ref<bool(AbstractCallSite)> Pred, 
-                             const AbstractAttribute &QueryingAA, 
-                             bool RequireAllCallSites, 
-                             bool &UsedAssumedInformation); 
-   
-   /// Check \p Pred on all call sites of \p Fn. 
-   /// 
-   /// This method will evaluate \p Pred on call sites and return 
-   /// true if \p Pred holds in every call sites. However, this is only possible 
-   /// all call sites are known, hence the function has internal linkage. 
-   /// If true is returned, \p UsedAssumedInformation is set if assumed 
-   /// information was used to skip or simplify potential call sites. 
-   bool checkForAllCallSites(function_ref<bool(AbstractCallSite)> Pred, 
-                             const Function &Fn, bool RequireAllCallSites, 
-                             const AbstractAttribute *QueryingAA, 
-                             bool &UsedAssumedInformation, 
-                             bool CheckPotentiallyDead = false); 
-   
-   /// Check \p Pred on all values potentially returned by \p F. 
-   /// 
-   /// This method will evaluate \p Pred on all values potentially returned by 
-   /// the function associated with \p QueryingAA. The returned values are 
-   /// matched with their respective return instructions. Returns true if \p Pred 
-   /// holds on all of them. 
-   bool checkForAllReturnedValuesAndReturnInsts( 
-       function_ref<bool(Value &, const SmallSetVector<ReturnInst *, 4> &)> Pred, 
-       const AbstractAttribute &QueryingAA); 
-   
-   /// Check \p Pred on all values potentially returned by the function 
-   /// associated with \p QueryingAA. 
-   /// 
-   /// This is the context insensitive version of the method above. 
-   bool checkForAllReturnedValues(function_ref<bool(Value &)> Pred, 
-                                  const AbstractAttribute &QueryingAA); 
-   
-   /// Check \p Pred on all instructions in \p Fn with an opcode present in 
-   /// \p Opcodes. 
-   /// 
-   /// This method will evaluate \p Pred on all instructions with an opcode 
-   /// present in \p Opcode and return true if \p Pred holds on all of them. 
-   bool checkForAllInstructions(function_ref<bool(Instruction &)> Pred, 
-                                const Function *Fn, 
-                                const AbstractAttribute &QueryingAA, 
-                                const ArrayRef<unsigned> &Opcodes, 
-                                bool &UsedAssumedInformation, 
-                                bool CheckBBLivenessOnly = false, 
-                                bool CheckPotentiallyDead = false); 
-   
-   /// Check \p Pred on all instructions with an opcode present in \p Opcodes. 
-   /// 
-   /// This method will evaluate \p Pred on all instructions with an opcode 
-   /// present in \p Opcode and return true if \p Pred holds on all of them. 
-   bool checkForAllInstructions(function_ref<bool(Instruction &)> Pred, 
-                                const AbstractAttribute &QueryingAA, 
-                                const ArrayRef<unsigned> &Opcodes, 
-                                bool &UsedAssumedInformation, 
-                                bool CheckBBLivenessOnly = false, 
-                                bool CheckPotentiallyDead = false); 
-   
-   /// Check \p Pred on all call-like instructions (=CallBased derived). 
-   /// 
-   /// See checkForAllCallLikeInstructions(...) for more information. 
-   bool checkForAllCallLikeInstructions(function_ref<bool(Instruction &)> Pred, 
-                                        const AbstractAttribute &QueryingAA, 
-                                        bool &UsedAssumedInformation, 
-                                        bool CheckBBLivenessOnly = false, 
-                                        bool CheckPotentiallyDead = false) { 
-     return checkForAllInstructions( 
-         Pred, QueryingAA, 
-         {(unsigned)Instruction::Invoke, (unsigned)Instruction::CallBr, 
-          (unsigned)Instruction::Call}, 
-         UsedAssumedInformation, CheckBBLivenessOnly, CheckPotentiallyDead); 
-   } 
-   
-   /// Check \p Pred on all Read/Write instructions. 
-   /// 
-   /// This method will evaluate \p Pred on all instructions that read or write 
-   /// to memory present in the information cache and return true if \p Pred 
-   /// holds on all of them. 
-   bool checkForAllReadWriteInstructions(function_ref<bool(Instruction &)> Pred, 
-                                         AbstractAttribute &QueryingAA, 
-                                         bool &UsedAssumedInformation); 
-   
-   /// Create a shallow wrapper for \p F such that \p F has internal linkage 
-   /// afterwards. It also sets the original \p F 's name to anonymous 
-   /// 
-   /// A wrapper is a function with the same type (and attributes) as \p F 
-   /// that will only call \p F and return the result, if any. 
-   /// 
-   /// Assuming the declaration of looks like: 
-   ///   rty F(aty0 arg0, ..., atyN argN); 
-   /// 
-   /// The wrapper will then look as follows: 
-   ///   rty wrapper(aty0 arg0, ..., atyN argN) { 
-   ///     return F(arg0, ..., argN); 
-   ///   } 
-   /// 
-   static void createShallowWrapper(Function &F); 
-   
-   /// Returns true if the function \p F can be internalized. i.e. it has a 
-   /// compatible linkage. 
-   static bool isInternalizable(Function &F); 
-   
-   /// Make another copy of the function \p F such that the copied version has 
-   /// internal linkage afterwards and can be analysed. Then we replace all uses 
-   /// of the original function to the copied one 
-   /// 
-   /// Only non-locally linked functions that have `linkonce_odr` or `weak_odr` 
-   /// linkage can be internalized because these linkages guarantee that other 
-   /// definitions with the same name have the same semantics as this one. 
-   /// 
-   /// This will only be run if the `attributor-allow-deep-wrappers` option is 
-   /// set, or if the function is called with \p Force set to true. 
-   /// 
-   /// If the function \p F failed to be internalized the return value will be a 
-   /// null pointer. 
-   static Function *internalizeFunction(Function &F, bool Force = false); 
-   
-   /// Make copies of each function in the set \p FnSet such that the copied 
-   /// version has internal linkage afterwards and can be analysed. Then we 
-   /// replace all uses of the original function to the copied one. The map 
-   /// \p FnMap contains a mapping of functions to their internalized versions. 
-   /// 
-   /// Only non-locally linked functions that have `linkonce_odr` or `weak_odr` 
-   /// linkage can be internalized because these linkages guarantee that other 
-   /// definitions with the same name have the same semantics as this one. 
-   /// 
-   /// This version will internalize all the functions in the set \p FnSet at 
-   /// once and then replace the uses. This prevents internalized functions being 
-   /// called by external functions when there is an internalized version in the 
-   /// module. 
-   static bool internalizeFunctions(SmallPtrSetImpl<Function *> &FnSet, 
-                                    DenseMap<Function *, Function *> &FnMap); 
-   
-   /// Return the data layout associated with the anchor scope. 
-   const DataLayout &getDataLayout() const { return InfoCache.DL; } 
-   
-   /// The allocator used to allocate memory, e.g. for `AbstractAttribute`s. 
-   BumpPtrAllocator &Allocator; 
-   
- private: 
-   /// This method will do fixpoint iteration until fixpoint or the 
-   /// maximum iteration count is reached. 
-   /// 
-   /// If the maximum iteration count is reached, This method will 
-   /// indicate pessimistic fixpoint on attributes that transitively depend 
-   /// on attributes that were scheduled for an update. 
-   void runTillFixpoint(); 
-   
-   /// Gets called after scheduling, manifests attributes to the LLVM IR. 
-   ChangeStatus manifestAttributes(); 
-   
-   /// Gets called after attributes have been manifested, cleans up the IR. 
-   /// Deletes dead functions, blocks and instructions. 
-   /// Rewrites function signitures and updates the call graph. 
-   ChangeStatus cleanupIR(); 
-   
-   /// Identify internal functions that are effectively dead, thus not reachable 
-   /// from a live entry point. The functions are added to ToBeDeletedFunctions. 
-   void identifyDeadInternalFunctions(); 
-   
-   /// Run `::update` on \p AA and track the dependences queried while doing so. 
-   /// Also adjust the state if we know further updates are not necessary. 
-   ChangeStatus updateAA(AbstractAttribute &AA); 
-   
-   /// Remember the dependences on the top of the dependence stack such that they 
-   /// may trigger further updates. (\see DependenceStack) 
-   void rememberDependences(); 
-   
-   /// Determine if CallBase context in \p IRP should be propagated. 
-   bool shouldPropagateCallBaseContext(const IRPosition &IRP); 
-   
-   /// Apply all requested function signature rewrites 
-   /// (\see registerFunctionSignatureRewrite) and return Changed if the module 
-   /// was altered. 
-   ChangeStatus 
-   rewriteFunctionSignatures(SmallSetVector<Function *, 8> &ModifiedFns); 
-   
-   /// Check if the Attribute \p AA should be seeded. 
-   /// See getOrCreateAAFor. 
-   bool shouldSeedAttribute(AbstractAttribute &AA); 
-   
-   /// A nested map to lookup abstract attributes based on the argument position 
-   /// on the outer level, and the addresses of the static member (AAType::ID) on 
-   /// the inner level. 
-   ///{ 
-   using AAMapKeyTy = std::pair<const char *, IRPosition>; 
-   DenseMap<AAMapKeyTy, AbstractAttribute *> AAMap; 
-   ///} 
-   
-   /// Map to remember all requested signature changes (= argument replacements). 
-   DenseMap<Function *, SmallVector<std::unique_ptr<ArgumentReplacementInfo>, 8>> 
-       ArgumentReplacementMap; 
-   
-   /// The set of functions we are deriving attributes for. 
-   SetVector<Function *> &Functions; 
-   
-   /// The information cache that holds pre-processed (LLVM-IR) information. 
-   InformationCache &InfoCache; 
-   
-   /// Abstract Attribute dependency graph 
-   AADepGraph DG; 
-   
-   /// Set of functions for which we modified the content such that it might 
-   /// impact the call graph. 
-   SmallSetVector<Function *, 8> CGModifiedFunctions; 
-   
-   /// Information about a dependence. If FromAA is changed ToAA needs to be 
-   /// updated as well. 
-   struct DepInfo { 
-     const AbstractAttribute *FromAA; 
-     const AbstractAttribute *ToAA; 
-     DepClassTy DepClass; 
-   }; 
-   
-   /// The dependence stack is used to track dependences during an 
-   /// `AbstractAttribute::update` call. As `AbstractAttribute::update` can be 
-   /// recursive we might have multiple vectors of dependences in here. The stack 
-   /// size, should be adjusted according to the expected recursion depth and the 
-   /// inner dependence vector size to the expected number of dependences per 
-   /// abstract attribute. Since the inner vectors are actually allocated on the 
-   /// stack we can be generous with their size. 
-   using DependenceVector = SmallVector<DepInfo, 8>; 
-   SmallVector<DependenceVector *, 16> DependenceStack; 
-   
-   /// A set to remember the functions we already assume to be live and visited. 
-   DenseSet<const Function *> VisitedFunctions; 
-   
-   /// Uses we replace with a new value after manifest is done. We will remove 
-   /// then trivially dead instructions as well. 
-   SmallMapVector<Use *, Value *, 32> ToBeChangedUses; 
-   
-   /// Values we replace with a new value after manifest is done. We will remove 
-   /// then trivially dead instructions as well. 
-   SmallMapVector<Value *, PointerIntPair<Value *, 1, bool>, 32> 
-       ToBeChangedValues; 
-   
-   /// Instructions we replace with `unreachable` insts after manifest is done. 
-   SmallSetVector<WeakVH, 16> ToBeChangedToUnreachableInsts; 
-   
-   /// Invoke instructions with at least a single dead successor block. 
-   SmallSetVector<WeakVH, 16> InvokeWithDeadSuccessor; 
-   
-   /// A flag that indicates which stage of the process we are in. Initially, the 
-   /// phase is SEEDING. Phase is changed in `Attributor::run()` 
-   enum class AttributorPhase { 
-     SEEDING, 
-     UPDATE, 
-     MANIFEST, 
-     CLEANUP, 
-   } Phase = AttributorPhase::SEEDING; 
-   
-   /// The current initialization chain length. Tracked to avoid stack overflows. 
-   unsigned InitializationChainLength = 0; 
-   
-   /// Functions, blocks, and instructions we delete after manifest is done. 
-   /// 
-   ///{ 
-   SmallPtrSet<BasicBlock *, 8> ManifestAddedBlocks; 
-   SmallSetVector<Function *, 8> ToBeDeletedFunctions; 
-   SmallSetVector<BasicBlock *, 8> ToBeDeletedBlocks; 
-   SmallSetVector<WeakVH, 8> ToBeDeletedInsts; 
-   ///} 
-   
-   /// Container with all the query AAs that requested an update via 
-   /// registerForUpdate. 
-   SmallSetVector<AbstractAttribute *, 16> QueryAAsAwaitingUpdate; 
-   
-   /// User provided configuration for this Attributor instance. 
-   const AttributorConfig Configuration; 
-   
-   friend AADepGraph; 
-   friend AttributorCallGraph; 
- }; 
-   
- /// An interface to query the internal state of an abstract attribute. 
- /// 
- /// The abstract state is a minimal interface that allows the Attributor to 
- /// communicate with the abstract attributes about their internal state without 
- /// enforcing or exposing implementation details, e.g., the (existence of an) 
- /// underlying lattice. 
- /// 
- /// It is sufficient to be able to query if a state is (1) valid or invalid, (2) 
- /// at a fixpoint, and to indicate to the state that (3) an optimistic fixpoint 
- /// was reached or (4) a pessimistic fixpoint was enforced. 
- /// 
- /// All methods need to be implemented by the subclass. For the common use case, 
- /// a single boolean state or a bit-encoded state, the BooleanState and 
- /// {Inc,Dec,Bit}IntegerState classes are already provided. An abstract 
- /// attribute can inherit from them to get the abstract state interface and 
- /// additional methods to directly modify the state based if needed. See the 
- /// class comments for help. 
- struct AbstractState { 
-   virtual ~AbstractState() = default; 
-   
-   /// Return if this abstract state is in a valid state. If false, no 
-   /// information provided should be used. 
-   virtual bool isValidState() const = 0; 
-   
-   /// Return if this abstract state is fixed, thus does not need to be updated 
-   /// if information changes as it cannot change itself. 
-   virtual bool isAtFixpoint() const = 0; 
-   
-   /// Indicate that the abstract state should converge to the optimistic state. 
-   /// 
-   /// This will usually make the optimistically assumed state the known to be 
-   /// true state. 
-   /// 
-   /// \returns ChangeStatus::UNCHANGED as the assumed value should not change. 
-   virtual ChangeStatus indicateOptimisticFixpoint() = 0; 
-   
-   /// Indicate that the abstract state should converge to the pessimistic state. 
-   /// 
-   /// This will usually revert the optimistically assumed state to the known to 
-   /// be true state. 
-   /// 
-   /// \returns ChangeStatus::CHANGED as the assumed value may change. 
-   virtual ChangeStatus indicatePessimisticFixpoint() = 0; 
- }; 
-   
- /// Simple state with integers encoding. 
- /// 
- /// The interface ensures that the assumed bits are always a subset of the known 
- /// bits. Users can only add known bits and, except through adding known bits, 
- /// they can only remove assumed bits. This should guarantee monotoniticy and 
- /// thereby the existence of a fixpoint (if used corretly). The fixpoint is 
- /// reached when the assumed and known state/bits are equal. Users can 
- /// force/inidicate a fixpoint. If an optimistic one is indicated, the known 
- /// state will catch up with the assumed one, for a pessimistic fixpoint it is 
- /// the other way around. 
- template <typename base_ty, base_ty BestState, base_ty WorstState> 
- struct IntegerStateBase : public AbstractState { 
-   using base_t = base_ty; 
-   
-   IntegerStateBase() = default; 
-   IntegerStateBase(base_t Assumed) : Assumed(Assumed) {} 
-   
-   /// Return the best possible representable state. 
-   static constexpr base_t getBestState() { return BestState; } 
-   static constexpr base_t getBestState(const IntegerStateBase &) { 
-     return getBestState(); 
-   } 
-   
-   /// Return the worst possible representable state. 
-   static constexpr base_t getWorstState() { return WorstState; } 
-   static constexpr base_t getWorstState(const IntegerStateBase &) { 
-     return getWorstState(); 
-   } 
-   
-   /// See AbstractState::isValidState() 
-   /// NOTE: For now we simply pretend that the worst possible state is invalid. 
-   bool isValidState() const override { return Assumed != getWorstState(); } 
-   
-   /// See AbstractState::isAtFixpoint() 
-   bool isAtFixpoint() const override { return Assumed == Known; } 
-   
-   /// See AbstractState::indicateOptimisticFixpoint(...) 
-   ChangeStatus indicateOptimisticFixpoint() override { 
-     Known = Assumed; 
-     return ChangeStatus::UNCHANGED; 
-   } 
-   
-   /// See AbstractState::indicatePessimisticFixpoint(...) 
-   ChangeStatus indicatePessimisticFixpoint() override { 
-     Assumed = Known; 
-     return ChangeStatus::CHANGED; 
-   } 
-   
-   /// Return the known state encoding 
-   base_t getKnown() const { return Known; } 
-   
-   /// Return the assumed state encoding. 
-   base_t getAssumed() const { return Assumed; } 
-   
-   /// Equality for IntegerStateBase. 
-   bool 
-   operator==(const IntegerStateBase<base_t, BestState, WorstState> &R) const { 
-     return this->getAssumed() == R.getAssumed() && 
-            this->getKnown() == R.getKnown(); 
-   } 
-   
-   /// Inequality for IntegerStateBase. 
-   bool 
-   operator!=(const IntegerStateBase<base_t, BestState, WorstState> &R) const { 
-     return !(*this == R); 
-   } 
-   
-   /// "Clamp" this state with \p R. The result is subtype dependent but it is 
-   /// intended that only information assumed in both states will be assumed in 
-   /// this one afterwards. 
-   void operator^=(const IntegerStateBase<base_t, BestState, WorstState> &R) { 
-     handleNewAssumedValue(R.getAssumed()); 
-   } 
-   
-   /// "Clamp" this state with \p R. The result is subtype dependent but it is 
-   /// intended that information known in either state will be known in 
-   /// this one afterwards. 
-   void operator+=(const IntegerStateBase<base_t, BestState, WorstState> &R) { 
-     handleNewKnownValue(R.getKnown()); 
-   } 
-   
-   void operator|=(const IntegerStateBase<base_t, BestState, WorstState> &R) { 
-     joinOR(R.getAssumed(), R.getKnown()); 
-   } 
-   
-   void operator&=(const IntegerStateBase<base_t, BestState, WorstState> &R) { 
-     joinAND(R.getAssumed(), R.getKnown()); 
-   } 
-   
- protected: 
-   /// Handle a new assumed value \p Value. Subtype dependent. 
-   virtual void handleNewAssumedValue(base_t Value) = 0; 
-   
-   /// Handle a new known value \p Value. Subtype dependent. 
-   virtual void handleNewKnownValue(base_t Value) = 0; 
-   
-   /// Handle a  value \p Value. Subtype dependent. 
-   virtual void joinOR(base_t AssumedValue, base_t KnownValue) = 0; 
-   
-   /// Handle a new assumed value \p Value. Subtype dependent. 
-   virtual void joinAND(base_t AssumedValue, base_t KnownValue) = 0; 
-   
-   /// The known state encoding in an integer of type base_t. 
-   base_t Known = getWorstState(); 
-   
-   /// The assumed state encoding in an integer of type base_t. 
-   base_t Assumed = getBestState(); 
- }; 
-   
- /// Specialization of the integer state for a bit-wise encoding. 
- template <typename base_ty = uint32_t, base_ty BestState = ~base_ty(0), 
-           base_ty WorstState = 0> 
- struct BitIntegerState 
-     : public IntegerStateBase<base_ty, BestState, WorstState> { 
-   using base_t = base_ty; 
-   
-   /// Return true if the bits set in \p BitsEncoding are "known bits". 
-   bool isKnown(base_t BitsEncoding) const { 
-     return (this->Known & BitsEncoding) == BitsEncoding; 
-   } 
-   
-   /// Return true if the bits set in \p BitsEncoding are "assumed bits". 
-   bool isAssumed(base_t BitsEncoding) const { 
-     return (this->Assumed & BitsEncoding) == BitsEncoding; 
-   } 
-   
-   /// Add the bits in \p BitsEncoding to the "known bits". 
-   BitIntegerState &addKnownBits(base_t Bits) { 
-     // Make sure we never miss any "known bits". 
-     this->Assumed |= Bits; 
-     this->Known |= Bits; 
-     return *this; 
-   } 
-   
-   /// Remove the bits in \p BitsEncoding from the "assumed bits" if not known. 
-   BitIntegerState &removeAssumedBits(base_t BitsEncoding) { 
-     return intersectAssumedBits(~BitsEncoding); 
-   } 
-   
-   /// Remove the bits in \p BitsEncoding from the "known bits". 
-   BitIntegerState &removeKnownBits(base_t BitsEncoding) { 
-     this->Known = (this->Known & ~BitsEncoding); 
-     return *this; 
-   } 
-   
-   /// Keep only "assumed bits" also set in \p BitsEncoding but all known ones. 
-   BitIntegerState &intersectAssumedBits(base_t BitsEncoding) { 
-     // Make sure we never loose any "known bits". 
-     this->Assumed = (this->Assumed & BitsEncoding) | this->Known; 
-     return *this; 
-   } 
-   
- private: 
-   void handleNewAssumedValue(base_t Value) override { 
-     intersectAssumedBits(Value); 
-   } 
-   void handleNewKnownValue(base_t Value) override { addKnownBits(Value); } 
-   void joinOR(base_t AssumedValue, base_t KnownValue) override { 
-     this->Known |= KnownValue; 
-     this->Assumed |= AssumedValue; 
-   } 
-   void joinAND(base_t AssumedValue, base_t KnownValue) override { 
-     this->Known &= KnownValue; 
-     this->Assumed &= AssumedValue; 
-   } 
- }; 
-   
- /// Specialization of the integer state for an increasing value, hence ~0u is 
- /// the best state and 0 the worst. 
- template <typename base_ty = uint32_t, base_ty BestState = ~base_ty(0), 
-           base_ty WorstState = 0> 
- struct IncIntegerState 
-     : public IntegerStateBase<base_ty, BestState, WorstState> { 
-   using super = IntegerStateBase<base_ty, BestState, WorstState>; 
-   using base_t = base_ty; 
-   
-   IncIntegerState() : super() {} 
-   IncIntegerState(base_t Assumed) : super(Assumed) {} 
-   
-   /// Return the best possible representable state. 
-   static constexpr base_t getBestState() { return BestState; } 
-   static constexpr base_t 
-   getBestState(const IncIntegerState<base_ty, BestState, WorstState> &) { 
-     return getBestState(); 
-   } 
-   
-   /// Take minimum of assumed and \p Value. 
-   IncIntegerState &takeAssumedMinimum(base_t Value) { 
-     // Make sure we never loose "known value". 
-     this->Assumed = std::max(std::min(this->Assumed, Value), this->Known); 
-     return *this; 
-   } 
-   
-   /// Take maximum of known and \p Value. 
-   IncIntegerState &takeKnownMaximum(base_t Value) { 
-     // Make sure we never loose "known value". 
-     this->Assumed = std::max(Value, this->Assumed); 
-     this->Known = std::max(Value, this->Known); 
-     return *this; 
-   } 
-   
- private: 
-   void handleNewAssumedValue(base_t Value) override { 
-     takeAssumedMinimum(Value); 
-   } 
-   void handleNewKnownValue(base_t Value) override { takeKnownMaximum(Value); } 
-   void joinOR(base_t AssumedValue, base_t KnownValue) override { 
-     this->Known = std::max(this->Known, KnownValue); 
-     this->Assumed = std::max(this->Assumed, AssumedValue); 
-   } 
-   void joinAND(base_t AssumedValue, base_t KnownValue) override { 
-     this->Known = std::min(this->Known, KnownValue); 
-     this->Assumed = std::min(this->Assumed, AssumedValue); 
-   } 
- }; 
-   
- /// Specialization of the integer state for a decreasing value, hence 0 is the 
- /// best state and ~0u the worst. 
- template <typename base_ty = uint32_t> 
- struct DecIntegerState : public IntegerStateBase<base_ty, 0, ~base_ty(0)> { 
-   using base_t = base_ty; 
-   
-   /// Take maximum of assumed and \p Value. 
-   DecIntegerState &takeAssumedMaximum(base_t Value) { 
-     // Make sure we never loose "known value". 
-     this->Assumed = std::min(std::max(this->Assumed, Value), this->Known); 
-     return *this; 
-   } 
-   
-   /// Take minimum of known and \p Value. 
-   DecIntegerState &takeKnownMinimum(base_t Value) { 
-     // Make sure we never loose "known value". 
-     this->Assumed = std::min(Value, this->Assumed); 
-     this->Known = std::min(Value, this->Known); 
-     return *this; 
-   } 
-   
- private: 
-   void handleNewAssumedValue(base_t Value) override { 
-     takeAssumedMaximum(Value); 
-   } 
-   void handleNewKnownValue(base_t Value) override { takeKnownMinimum(Value); } 
-   void joinOR(base_t AssumedValue, base_t KnownValue) override { 
-     this->Assumed = std::min(this->Assumed, KnownValue); 
-     this->Assumed = std::min(this->Assumed, AssumedValue); 
-   } 
-   void joinAND(base_t AssumedValue, base_t KnownValue) override { 
-     this->Assumed = std::max(this->Assumed, KnownValue); 
-     this->Assumed = std::max(this->Assumed, AssumedValue); 
-   } 
- }; 
-   
- /// Simple wrapper for a single bit (boolean) state. 
- struct BooleanState : public IntegerStateBase<bool, true, false> { 
-   using super = IntegerStateBase<bool, true, false>; 
-   using base_t = IntegerStateBase::base_t; 
-   
-   BooleanState() = default; 
-   BooleanState(base_t Assumed) : super(Assumed) {} 
-   
-   /// Set the assumed value to \p Value but never below the known one. 
-   void setAssumed(bool Value) { Assumed &= (Known | Value); } 
-   
-   /// Set the known and asssumed value to \p Value. 
-   void setKnown(bool Value) { 
-     Known |= Value; 
-     Assumed |= Value; 
-   } 
-   
-   /// Return true if the state is assumed to hold. 
-   bool isAssumed() const { return getAssumed(); } 
-   
-   /// Return true if the state is known to hold. 
-   bool isKnown() const { return getKnown(); } 
-   
- private: 
-   void handleNewAssumedValue(base_t Value) override { 
-     if (!Value) 
-       Assumed = Known; 
-   } 
-   void handleNewKnownValue(base_t Value) override { 
-     if (Value) 
-       Known = (Assumed = Value); 
-   } 
-   void joinOR(base_t AssumedValue, base_t KnownValue) override { 
-     Known |= KnownValue; 
-     Assumed |= AssumedValue; 
-   } 
-   void joinAND(base_t AssumedValue, base_t KnownValue) override { 
-     Known &= KnownValue; 
-     Assumed &= AssumedValue; 
-   } 
- }; 
-   
- /// State for an integer range. 
- struct IntegerRangeState : public AbstractState { 
-   
-   /// Bitwidth of the associated value. 
-   uint32_t BitWidth; 
-   
-   /// State representing assumed range, initially set to empty. 
-   ConstantRange Assumed; 
-   
-   /// State representing known range, initially set to [-inf, inf]. 
-   ConstantRange Known; 
-   
-   IntegerRangeState(uint32_t BitWidth) 
-       : BitWidth(BitWidth), Assumed(ConstantRange::getEmpty(BitWidth)), 
-         Known(ConstantRange::getFull(BitWidth)) {} 
-   
-   IntegerRangeState(const ConstantRange &CR) 
-       : BitWidth(CR.getBitWidth()), Assumed(CR), 
-         Known(getWorstState(CR.getBitWidth())) {} 
-   
-   /// Return the worst possible representable state. 
-   static ConstantRange getWorstState(uint32_t BitWidth) { 
-     return ConstantRange::getFull(BitWidth); 
-   } 
-   
-   /// Return the best possible representable state. 
-   static ConstantRange getBestState(uint32_t BitWidth) { 
-     return ConstantRange::getEmpty(BitWidth); 
-   } 
-   static ConstantRange getBestState(const IntegerRangeState &IRS) { 
-     return getBestState(IRS.getBitWidth()); 
-   } 
-   
-   /// Return associated values' bit width. 
-   uint32_t getBitWidth() const { return BitWidth; } 
-   
-   /// See AbstractState::isValidState() 
-   bool isValidState() const override { 
-     return BitWidth > 0 && !Assumed.isFullSet(); 
-   } 
-   
-   /// See AbstractState::isAtFixpoint() 
-   bool isAtFixpoint() const override { return Assumed == Known; } 
-   
-   /// See AbstractState::indicateOptimisticFixpoint(...) 
-   ChangeStatus indicateOptimisticFixpoint() override { 
-     Known = Assumed; 
-     return ChangeStatus::CHANGED; 
-   } 
-   
-   /// See AbstractState::indicatePessimisticFixpoint(...) 
-   ChangeStatus indicatePessimisticFixpoint() override { 
-     Assumed = Known; 
-     return ChangeStatus::CHANGED; 
-   } 
-   
-   /// Return the known state encoding 
-   ConstantRange getKnown() const { return Known; } 
-   
-   /// Return the assumed state encoding. 
-   ConstantRange getAssumed() const { return Assumed; } 
-   
-   /// Unite assumed range with the passed state. 
-   void unionAssumed(const ConstantRange &R) { 
-     // Don't loose a known range. 
-     Assumed = Assumed.unionWith(R).intersectWith(Known); 
-   } 
-   
-   /// See IntegerRangeState::unionAssumed(..). 
-   void unionAssumed(const IntegerRangeState &R) { 
-     unionAssumed(R.getAssumed()); 
-   } 
-   
-   /// Intersect known range with the passed state. 
-   void intersectKnown(const ConstantRange &R) { 
-     Assumed = Assumed.intersectWith(R); 
-     Known = Known.intersectWith(R); 
-   } 
-   
-   /// See IntegerRangeState::intersectKnown(..). 
-   void intersectKnown(const IntegerRangeState &R) { 
-     intersectKnown(R.getKnown()); 
-   } 
-   
-   /// Equality for IntegerRangeState. 
-   bool operator==(const IntegerRangeState &R) const { 
-     return getAssumed() == R.getAssumed() && getKnown() == R.getKnown(); 
-   } 
-   
-   /// "Clamp" this state with \p R. The result is subtype dependent but it is 
-   /// intended that only information assumed in both states will be assumed in 
-   /// this one afterwards. 
-   IntegerRangeState operator^=(const IntegerRangeState &R) { 
-     // NOTE: `^=` operator seems like `intersect` but in this case, we need to 
-     // take `union`. 
-     unionAssumed(R); 
-     return *this; 
-   } 
-   
-   IntegerRangeState operator&=(const IntegerRangeState &R) { 
-     // NOTE: `&=` operator seems like `intersect` but in this case, we need to 
-     // take `union`. 
-     Known = Known.unionWith(R.getKnown()); 
-     Assumed = Assumed.unionWith(R.getAssumed()); 
-     return *this; 
-   } 
- }; 
-   
- /// Simple state for a set. 
- /// 
- /// This represents a state containing a set of values. The interface supports 
- /// modelling sets that contain all possible elements. The state's internal 
- /// value is modified using union or intersection operations. 
- template <typename BaseTy> struct SetState : public AbstractState { 
-   /// A wrapper around a set that has semantics for handling unions and 
-   /// intersections with a "universal" set that contains all elements. 
-   struct SetContents { 
-     /// Creates a universal set with no concrete elements or an empty set. 
-     SetContents(bool Universal) : Universal(Universal) {} 
-   
-     /// Creates a non-universal set with concrete values. 
-     SetContents(const DenseSet<BaseTy> &Assumptions) 
-         : Universal(false), Set(Assumptions) {} 
-   
-     SetContents(bool Universal, const DenseSet<BaseTy> &Assumptions) 
-         : Universal(Universal), Set(Assumptions) {} 
-   
-     const DenseSet<BaseTy> &getSet() const { return Set; } 
-   
-     bool isUniversal() const { return Universal; } 
-   
-     bool empty() const { return Set.empty() && !Universal; } 
-   
-     /// Finds A := A ^ B where A or B could be the "Universal" set which 
-     /// contains every possible attribute. Returns true if changes were made. 
-     bool getIntersection(const SetContents &RHS) { 
-       bool IsUniversal = Universal; 
-       unsigned Size = Set.size(); 
-   
-       // A := A ^ U = A 
-       if (RHS.isUniversal()) 
-         return false; 
-   
-       // A := U ^ B = B 
-       if (Universal) 
-         Set = RHS.getSet(); 
-       else 
-         set_intersect(Set, RHS.getSet()); 
-   
-       Universal &= RHS.isUniversal(); 
-       return IsUniversal != Universal || Size != Set.size(); 
-     } 
-   
-     /// Finds A := A u B where A or B could be the "Universal" set which 
-     /// contains every possible attribute. returns true if changes were made. 
-     bool getUnion(const SetContents &RHS) { 
-       bool IsUniversal = Universal; 
-       unsigned Size = Set.size(); 
-   
-       // A := A u U = U = U u B 
-       if (!RHS.isUniversal() && !Universal) 
-         set_union(Set, RHS.getSet()); 
-   
-       Universal |= RHS.isUniversal(); 
-       return IsUniversal != Universal || Size != Set.size(); 
-     } 
-   
-   private: 
-     /// Indicates if this set is "universal", containing every possible element. 
-     bool Universal; 
-   
-     /// The set of currently active assumptions. 
-     DenseSet<BaseTy> Set; 
-   }; 
-   
-   SetState() : Known(false), Assumed(true), IsAtFixedpoint(false) {} 
-   
-   /// Initializes the known state with an initial set and initializes the 
-   /// assumed state as universal. 
-   SetState(const DenseSet<BaseTy> &Known) 
-       : Known(Known), Assumed(true), IsAtFixedpoint(false) {} 
-   
-   /// See AbstractState::isValidState() 
-   bool isValidState() const override { return !Assumed.empty(); } 
-   
-   /// See AbstractState::isAtFixpoint() 
-   bool isAtFixpoint() const override { return IsAtFixedpoint; } 
-   
-   /// See AbstractState::indicateOptimisticFixpoint(...) 
-   ChangeStatus indicateOptimisticFixpoint() override { 
-     IsAtFixedpoint = true; 
-     Known = Assumed; 
-     return ChangeStatus::UNCHANGED; 
-   } 
-   
-   /// See AbstractState::indicatePessimisticFixpoint(...) 
-   ChangeStatus indicatePessimisticFixpoint() override { 
-     IsAtFixedpoint = true; 
-     Assumed = Known; 
-     return ChangeStatus::CHANGED; 
-   } 
-   
-   /// Return the known state encoding. 
-   const SetContents &getKnown() const { return Known; } 
-   
-   /// Return the assumed state encoding. 
-   const SetContents &getAssumed() const { return Assumed; } 
-   
-   /// Returns if the set state contains the element. 
-   bool setContains(const BaseTy &Elem) const { 
-     return Assumed.getSet().contains(Elem) || Known.getSet().contains(Elem); 
-   } 
-   
-   /// Performs the set intersection between this set and \p RHS. Returns true if 
-   /// changes were made. 
-   bool getIntersection(const SetContents &RHS) { 
-     unsigned SizeBefore = Assumed.getSet().size(); 
-   
-     // Get intersection and make sure that the known set is still a proper 
-     // subset of the assumed set. A := K u (A ^ R). 
-     Assumed.getIntersection(RHS); 
-     Assumed.getUnion(Known); 
-   
-     return SizeBefore != Assumed.getSet().size(); 
-   } 
-   
-   /// Performs the set union between this set and \p RHS. Returns true if 
-   /// changes were made. 
-   bool getUnion(const SetContents &RHS) { return Assumed.getUnion(RHS); } 
-   
- private: 
-   /// The set of values known for this state. 
-   SetContents Known; 
-   
-   /// The set of assumed values for this state. 
-   SetContents Assumed; 
-   
-   bool IsAtFixedpoint; 
- }; 
-   
- /// Helper struct necessary as the modular build fails if the virtual method 
- /// IRAttribute::manifest is defined in the Attributor.cpp. 
- struct IRAttributeManifest { 
-   static ChangeStatus manifestAttrs(Attributor &A, const IRPosition &IRP, 
-                                     const ArrayRef<Attribute> &DeducedAttrs, 
-                                     bool ForceReplace = false); 
- }; 
-   
- /// Helper to tie a abstract state implementation to an abstract attribute. 
- template <typename StateTy, typename BaseType, class... Ts> 
- struct StateWrapper : public BaseType, public StateTy { 
-   /// Provide static access to the type of the state. 
-   using StateType = StateTy; 
-   
-   StateWrapper(const IRPosition &IRP, Ts... Args) 
-       : BaseType(IRP), StateTy(Args...) {} 
-   
-   /// See AbstractAttribute::getState(...). 
-   StateType &getState() override { return *this; } 
-   
-   /// See AbstractAttribute::getState(...). 
-   const StateType &getState() const override { return *this; } 
- }; 
-   
- /// Helper class that provides common functionality to manifest IR attributes. 
- template <Attribute::AttrKind AK, typename BaseType> 
- struct IRAttribute : public BaseType { 
-   IRAttribute(const IRPosition &IRP) : BaseType(IRP) {} 
-   
-   /// See AbstractAttribute::initialize(...). 
-   void initialize(Attributor &A) override { 
-     const IRPosition &IRP = this->getIRPosition(); 
-     if (isa<UndefValue>(IRP.getAssociatedValue()) || 
-         this->hasAttr(getAttrKind(), /* IgnoreSubsumingPositions */ false, 
-                       &A)) { 
-       this->getState().indicateOptimisticFixpoint(); 
-       return; 
-     } 
-   
-     bool IsFnInterface = IRP.isFnInterfaceKind(); 
-     const Function *FnScope = IRP.getAnchorScope(); 
-     // TODO: Not all attributes require an exact definition. Find a way to 
-     //       enable deduction for some but not all attributes in case the 
-     //       definition might be changed at runtime, see also 
-     //       http://lists.llvm.org/pipermail/llvm-dev/2018-February/121275.html. 
-     // TODO: We could always determine abstract attributes and if sufficient 
-     //       information was found we could duplicate the functions that do not 
-     //       have an exact definition. 
-     if (IsFnInterface && (!FnScope || !A.isFunctionIPOAmendable(*FnScope))) 
-       this->getState().indicatePessimisticFixpoint(); 
-   } 
-   
-   /// See AbstractAttribute::manifest(...). 
-   ChangeStatus manifest(Attributor &A) override { 
-     if (isa<UndefValue>(this->getIRPosition().getAssociatedValue())) 
-       return ChangeStatus::UNCHANGED; 
-     SmallVector<Attribute, 4> DeducedAttrs; 
-     getDeducedAttributes(this->getAnchorValue().getContext(), DeducedAttrs); 
-     return IRAttributeManifest::manifestAttrs(A, this->getIRPosition(), 
-                                               DeducedAttrs); 
-   } 
-   
-   /// Return the kind that identifies the abstract attribute implementation. 
-   Attribute::AttrKind getAttrKind() const { return AK; } 
-   
-   /// Return the deduced attributes in \p Attrs. 
-   virtual void getDeducedAttributes(LLVMContext &Ctx, 
-                                     SmallVectorImpl<Attribute> &Attrs) const { 
-     Attrs.emplace_back(Attribute::get(Ctx, getAttrKind())); 
-   } 
- }; 
-   
- /// Base struct for all "concrete attribute" deductions. 
- /// 
- /// The abstract attribute is a minimal interface that allows the Attributor to 
- /// orchestrate the abstract/fixpoint analysis. The design allows to hide away 
- /// implementation choices made for the subclasses but also to structure their 
- /// implementation and simplify the use of other abstract attributes in-flight. 
- /// 
- /// To allow easy creation of new attributes, most methods have default 
- /// implementations. The ones that do not are generally straight forward, except 
- /// `AbstractAttribute::updateImpl` which is the location of most reasoning 
- /// associated with the abstract attribute. The update is invoked by the 
- /// Attributor in case the situation used to justify the current optimistic 
- /// state might have changed. The Attributor determines this automatically 
- /// by monitoring the `Attributor::getAAFor` calls made by abstract attributes. 
- /// 
- /// The `updateImpl` method should inspect the IR and other abstract attributes 
- /// in-flight to justify the best possible (=optimistic) state. The actual 
- /// implementation is, similar to the underlying abstract state encoding, not 
- /// exposed. In the most common case, the `updateImpl` will go through a list of 
- /// reasons why its optimistic state is valid given the current information. If 
- /// any combination of them holds and is sufficient to justify the current 
- /// optimistic state, the method shall return UNCHAGED. If not, the optimistic 
- /// state is adjusted to the situation and the method shall return CHANGED. 
- /// 
- /// If the manifestation of the "concrete attribute" deduced by the subclass 
- /// differs from the "default" behavior, which is a (set of) LLVM-IR 
- /// attribute(s) for an argument, call site argument, function return value, or 
- /// function, the `AbstractAttribute::manifest` method should be overloaded. 
- /// 
- /// NOTE: If the state obtained via getState() is INVALID, thus if 
- ///       AbstractAttribute::getState().isValidState() returns false, no 
- ///       information provided by the methods of this class should be used. 
- /// NOTE: The Attributor currently has certain limitations to what we can do. 
- ///       As a general rule of thumb, "concrete" abstract attributes should *for 
- ///       now* only perform "backward" information propagation. That means 
- ///       optimistic information obtained through abstract attributes should 
- ///       only be used at positions that precede the origin of the information 
- ///       with regards to the program flow. More practically, information can 
- ///       *now* be propagated from instructions to their enclosing function, but 
- ///       *not* from call sites to the called function. The mechanisms to allow 
- ///       both directions will be added in the future. 
- /// NOTE: The mechanics of adding a new "concrete" abstract attribute are 
- ///       described in the file comment. 
- struct AbstractAttribute : public IRPosition, public AADepGraphNode { 
-   using StateType = AbstractState; 
-   
-   AbstractAttribute(const IRPosition &IRP) : IRPosition(IRP) {} 
-   
-   /// Virtual destructor. 
-   virtual ~AbstractAttribute() = default; 
-   
-   /// This function is used to identify if an \p DGN is of type 
-   /// AbstractAttribute so that the dyn_cast and cast can use such information 
-   /// to cast an AADepGraphNode to an AbstractAttribute. 
-   /// 
-   /// We eagerly return true here because all AADepGraphNodes except for the 
-   /// Synthethis Node are of type AbstractAttribute 
-   static bool classof(const AADepGraphNode *DGN) { return true; } 
-   
-   /// Initialize the state with the information in the Attributor \p A. 
-   /// 
-   /// This function is called by the Attributor once all abstract attributes 
-   /// have been identified. It can and shall be used for task like: 
-   ///  - identify existing knowledge in the IR and use it for the "known state" 
-   ///  - perform any work that is not going to change over time, e.g., determine 
-   ///    a subset of the IR, or attributes in-flight, that have to be looked at 
-   ///    in the `updateImpl` method. 
-   virtual void initialize(Attributor &A) {} 
-   
-   /// A query AA is always scheduled as long as we do updates because it does 
-   /// lazy computation that cannot be determined to be done from the outside. 
-   /// However, while query AAs will not be fixed if they do not have outstanding 
-   /// dependences, we will only schedule them like other AAs. If a query AA that 
-   /// received a new query it needs to request an update via 
-   /// `Attributor::requestUpdateForAA`. 
-   virtual bool isQueryAA() const { return false; } 
-   
-   /// Return the internal abstract state for inspection. 
-   virtual StateType &getState() = 0; 
-   virtual const StateType &getState() const = 0; 
-   
-   /// Return an IR position, see struct IRPosition. 
-   const IRPosition &getIRPosition() const { return *this; }; 
-   IRPosition &getIRPosition() { return *this; }; 
-   
-   /// Helper functions, for debug purposes only. 
-   ///{ 
-   void print(raw_ostream &OS) const override; 
-   virtual void printWithDeps(raw_ostream &OS) const; 
-   void dump() const { print(dbgs()); } 
-   
-   /// This function should return the "summarized" assumed state as string. 
-   virtual const std::string getAsStr() const = 0; 
-   
-   /// This function should return the name of the AbstractAttribute 
-   virtual const std::string getName() const = 0; 
-   
-   /// This function should return the address of the ID of the AbstractAttribute 
-   virtual const char *getIdAddr() const = 0; 
-   ///} 
-   
-   /// Allow the Attributor access to the protected methods. 
-   friend struct Attributor; 
-   
- protected: 
-   /// Hook for the Attributor to trigger an update of the internal state. 
-   /// 
-   /// If this attribute is already fixed, this method will return UNCHANGED, 
-   /// otherwise it delegates to `AbstractAttribute::updateImpl`. 
-   /// 
-   /// \Return CHANGED if the internal state changed, otherwise UNCHANGED. 
-   ChangeStatus update(Attributor &A); 
-   
-   /// Hook for the Attributor to trigger the manifestation of the information 
-   /// represented by the abstract attribute in the LLVM-IR. 
-   /// 
-   /// \Return CHANGED if the IR was altered, otherwise UNCHANGED. 
-   virtual ChangeStatus manifest(Attributor &A) { 
-     return ChangeStatus::UNCHANGED; 
-   } 
-   
-   /// Hook to enable custom statistic tracking, called after manifest that 
-   /// resulted in a change if statistics are enabled. 
-   /// 
-   /// We require subclasses to provide an implementation so we remember to 
-   /// add statistics for them. 
-   virtual void trackStatistics() const = 0; 
-   
-   /// The actual update/transfer function which has to be implemented by the 
-   /// derived classes. 
-   /// 
-   /// If it is called, the environment has changed and we have to determine if 
-   /// the current information is still valid or adjust it otherwise. 
-   /// 
-   /// \Return CHANGED if the internal state changed, otherwise UNCHANGED. 
-   virtual ChangeStatus updateImpl(Attributor &A) = 0; 
- }; 
-   
- /// Forward declarations of output streams for debug purposes. 
- /// 
- ///{ 
- raw_ostream &operator<<(raw_ostream &OS, const AbstractAttribute &AA); 
- raw_ostream &operator<<(raw_ostream &OS, ChangeStatus S); 
- raw_ostream &operator<<(raw_ostream &OS, IRPosition::Kind); 
- raw_ostream &operator<<(raw_ostream &OS, const IRPosition &); 
- raw_ostream &operator<<(raw_ostream &OS, const AbstractState &State); 
- template <typename base_ty, base_ty BestState, base_ty WorstState> 
- raw_ostream & 
- operator<<(raw_ostream &OS, 
-            const IntegerStateBase<base_ty, BestState, WorstState> &S) { 
-   return OS << "(" << S.getKnown() << "-" << S.getAssumed() << ")" 
-             << static_cast<const AbstractState &>(S); 
- } 
- raw_ostream &operator<<(raw_ostream &OS, const IntegerRangeState &State); 
- ///} 
-   
- struct AttributorPass : public PassInfoMixin<AttributorPass> { 
-   PreservedAnalyses run(Module &M, ModuleAnalysisManager &AM); 
- }; 
- struct AttributorCGSCCPass : public PassInfoMixin<AttributorCGSCCPass> { 
-   PreservedAnalyses run(LazyCallGraph::SCC &C, CGSCCAnalysisManager &AM, 
-                         LazyCallGraph &CG, CGSCCUpdateResult &UR); 
- }; 
-   
- Pass *createAttributorLegacyPass(); 
- Pass *createAttributorCGSCCLegacyPass(); 
-   
- /// Helper function to clamp a state \p S of type \p StateType with the 
- /// information in \p R and indicate/return if \p S did change (as-in update is 
- /// required to be run again). 
- template <typename StateType> 
- ChangeStatus clampStateAndIndicateChange(StateType &S, const StateType &R) { 
-   auto Assumed = S.getAssumed(); 
-   S ^= R; 
-   return Assumed == S.getAssumed() ? ChangeStatus::UNCHANGED 
-                                    : ChangeStatus::CHANGED; 
- } 
-   
- /// ---------------------------------------------------------------------------- 
- ///                       Abstract Attribute Classes 
- /// ---------------------------------------------------------------------------- 
-   
- /// An abstract attribute for the returned values of a function. 
- struct AAReturnedValues 
-     : public IRAttribute<Attribute::Returned, AbstractAttribute> { 
-   AAReturnedValues(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Check \p Pred on all returned values. 
-   /// 
-   /// This method will evaluate \p Pred on returned values and return 
-   /// true if (1) all returned values are known, and (2) \p Pred returned true 
-   /// for all returned values. 
-   /// 
-   /// Note: Unlike the Attributor::checkForAllReturnedValuesAndReturnInsts 
-   /// method, this one will not filter dead return instructions. 
-   virtual bool checkForAllReturnedValuesAndReturnInsts( 
-       function_ref<bool(Value &, const SmallSetVector<ReturnInst *, 4> &)> Pred) 
-       const = 0; 
-   
-   using iterator = 
-       MapVector<Value *, SmallSetVector<ReturnInst *, 4>>::iterator; 
-   using const_iterator = 
-       MapVector<Value *, SmallSetVector<ReturnInst *, 4>>::const_iterator; 
-   virtual llvm::iterator_range<iterator> returned_values() = 0; 
-   virtual llvm::iterator_range<const_iterator> returned_values() const = 0; 
-   
-   virtual size_t getNumReturnValues() const = 0; 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAReturnedValues &createForPosition(const IRPosition &IRP, 
-                                              Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAReturnedValues"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAReturnedValues 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- struct AANoUnwind 
-     : public IRAttribute<Attribute::NoUnwind, 
-                          StateWrapper<BooleanState, AbstractAttribute>> { 
-   AANoUnwind(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Returns true if nounwind is assumed. 
-   bool isAssumedNoUnwind() const { return getAssumed(); } 
-   
-   /// Returns true if nounwind is known. 
-   bool isKnownNoUnwind() const { return getKnown(); } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AANoUnwind &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AANoUnwind"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AANoUnwind 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- struct AANoSync 
-     : public IRAttribute<Attribute::NoSync, 
-                          StateWrapper<BooleanState, AbstractAttribute>> { 
-   AANoSync(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Returns true if "nosync" is assumed. 
-   bool isAssumedNoSync() const { return getAssumed(); } 
-   
-   /// Returns true if "nosync" is known. 
-   bool isKnownNoSync() const { return getKnown(); } 
-   
-   /// Helper function used to determine whether an instruction is non-relaxed 
-   /// atomic. In other words, if an atomic instruction does not have unordered 
-   /// or monotonic ordering 
-   static bool isNonRelaxedAtomic(const Instruction *I); 
-   
-   /// Helper function specific for intrinsics which are potentially volatile. 
-   static bool isNoSyncIntrinsic(const Instruction *I); 
-   
-   /// Helper function to determine if \p CB is an aligned (GPU) barrier. Aligned 
-   /// barriers have to be executed by all threads. The flag \p ExecutedAligned 
-   /// indicates if the call is executed by all threads in a (thread) block in an 
-   /// aligned way. If that is the case, non-aligned barriers are effectively 
-   /// aligned barriers. 
-   static bool isAlignedBarrier(const CallBase &CB, bool ExecutedAligned); 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AANoSync &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AANoSync"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AANoSync 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract interface for all nonnull attributes. 
- struct AANonNull 
-     : public IRAttribute<Attribute::NonNull, 
-                          StateWrapper<BooleanState, AbstractAttribute>> { 
-   AANonNull(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Return true if we assume that the underlying value is nonnull. 
-   bool isAssumedNonNull() const { return getAssumed(); } 
-   
-   /// Return true if we know that underlying value is nonnull. 
-   bool isKnownNonNull() const { return getKnown(); } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AANonNull &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AANonNull"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AANonNull 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract attribute for norecurse. 
- struct AANoRecurse 
-     : public IRAttribute<Attribute::NoRecurse, 
-                          StateWrapper<BooleanState, AbstractAttribute>> { 
-   AANoRecurse(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Return true if "norecurse" is assumed. 
-   bool isAssumedNoRecurse() const { return getAssumed(); } 
-   
-   /// Return true if "norecurse" is known. 
-   bool isKnownNoRecurse() const { return getKnown(); } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AANoRecurse &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AANoRecurse"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AANoRecurse 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract attribute for willreturn. 
- struct AAWillReturn 
-     : public IRAttribute<Attribute::WillReturn, 
-                          StateWrapper<BooleanState, AbstractAttribute>> { 
-   AAWillReturn(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Return true if "willreturn" is assumed. 
-   bool isAssumedWillReturn() const { return getAssumed(); } 
-   
-   /// Return true if "willreturn" is known. 
-   bool isKnownWillReturn() const { return getKnown(); } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAWillReturn &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAWillReturn"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AAWillReturn 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract attribute for undefined behavior. 
- struct AAUndefinedBehavior 
-     : public StateWrapper<BooleanState, AbstractAttribute> { 
-   using Base = StateWrapper<BooleanState, AbstractAttribute>; 
-   AAUndefinedBehavior(const IRPosition &IRP, Attributor &A) : Base(IRP) {} 
-   
-   /// Return true if "undefined behavior" is assumed. 
-   bool isAssumedToCauseUB() const { return getAssumed(); } 
-   
-   /// Return true if "undefined behavior" is assumed for a specific instruction. 
-   virtual bool isAssumedToCauseUB(Instruction *I) const = 0; 
-   
-   /// Return true if "undefined behavior" is known. 
-   bool isKnownToCauseUB() const { return getKnown(); } 
-   
-   /// Return true if "undefined behavior" is known for a specific instruction. 
-   virtual bool isKnownToCauseUB(Instruction *I) const = 0; 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAUndefinedBehavior &createForPosition(const IRPosition &IRP, 
-                                                 Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAUndefinedBehavior"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAUndefineBehavior 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract interface to determine reachability of point A to B. 
- struct AAIntraFnReachability 
-     : public StateWrapper<BooleanState, AbstractAttribute> { 
-   using Base = StateWrapper<BooleanState, AbstractAttribute>; 
-   AAIntraFnReachability(const IRPosition &IRP, Attributor &A) : Base(IRP) {} 
-   
-   /// Returns true if 'From' instruction is assumed to reach, 'To' instruction. 
-   /// Users should provide two positions they are interested in, and the class 
-   /// determines (and caches) reachability. 
-   virtual bool isAssumedReachable( 
-       Attributor &A, const Instruction &From, const Instruction &To, 
-       const AA::InstExclusionSetTy *ExclusionSet = nullptr) const = 0; 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAIntraFnReachability &createForPosition(const IRPosition &IRP, 
-                                                   Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAIntraFnReachability"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAIntraFnReachability 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract interface for all noalias attributes. 
- struct AANoAlias 
-     : public IRAttribute<Attribute::NoAlias, 
-                          StateWrapper<BooleanState, AbstractAttribute>> { 
-   AANoAlias(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Return true if we assume that the underlying value is alias. 
-   bool isAssumedNoAlias() const { return getAssumed(); } 
-   
-   /// Return true if we know that underlying value is noalias. 
-   bool isKnownNoAlias() const { return getKnown(); } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AANoAlias &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AANoAlias"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AANoAlias 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An AbstractAttribute for nofree. 
- struct AANoFree 
-     : public IRAttribute<Attribute::NoFree, 
-                          StateWrapper<BooleanState, AbstractAttribute>> { 
-   AANoFree(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Return true if "nofree" is assumed. 
-   bool isAssumedNoFree() const { return getAssumed(); } 
-   
-   /// Return true if "nofree" is known. 
-   bool isKnownNoFree() const { return getKnown(); } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AANoFree &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AANoFree"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AANoFree 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An AbstractAttribute for noreturn. 
- struct AANoReturn 
-     : public IRAttribute<Attribute::NoReturn, 
-                          StateWrapper<BooleanState, AbstractAttribute>> { 
-   AANoReturn(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Return true if the underlying object is assumed to never return. 
-   bool isAssumedNoReturn() const { return getAssumed(); } 
-   
-   /// Return true if the underlying object is known to never return. 
-   bool isKnownNoReturn() const { return getKnown(); } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AANoReturn &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AANoReturn"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AANoReturn 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract interface for liveness abstract attribute. 
- struct AAIsDead 
-     : public StateWrapper<BitIntegerState<uint8_t, 3, 0>, AbstractAttribute> { 
-   using Base = StateWrapper<BitIntegerState<uint8_t, 3, 0>, AbstractAttribute>; 
-   AAIsDead(const IRPosition &IRP, Attributor &A) : Base(IRP) {} 
-   
-   /// State encoding bits. A set bit in the state means the property holds. 
-   enum { 
-     HAS_NO_EFFECT = 1 << 0, 
-     IS_REMOVABLE = 1 << 1, 
-   
-     IS_DEAD = HAS_NO_EFFECT | IS_REMOVABLE, 
-   }; 
-   static_assert(IS_DEAD == getBestState(), "Unexpected BEST_STATE value"); 
-   
- protected: 
-   /// The query functions are protected such that other attributes need to go 
-   /// through the Attributor interfaces: `Attributor::isAssumedDead(...)` 
-   
-   /// Returns true if the underlying value is assumed dead. 
-   virtual bool isAssumedDead() const = 0; 
-   
-   /// Returns true if the underlying value is known dead. 
-   virtual bool isKnownDead() const = 0; 
-   
-   /// Returns true if \p BB is known dead. 
-   virtual bool isKnownDead(const BasicBlock *BB) const = 0; 
-   
-   /// Returns true if \p I is assumed dead. 
-   virtual bool isAssumedDead(const Instruction *I) const = 0; 
-   
-   /// Returns true if \p I is known dead. 
-   virtual bool isKnownDead(const Instruction *I) const = 0; 
-   
-   /// Return true if the underlying value is a store that is known to be 
-   /// removable. This is different from dead stores as the removable store 
-   /// can have an effect on live values, especially loads, but that effect 
-   /// is propagated which allows us to remove the store in turn. 
-   virtual bool isRemovableStore() const { return false; } 
-   
-   /// This method is used to check if at least one instruction in a collection 
-   /// of instructions is live. 
-   template <typename T> bool isLiveInstSet(T begin, T end) const { 
-     for (const auto &I : llvm::make_range(begin, end)) { 
-       assert(I->getFunction() == getIRPosition().getAssociatedFunction() && 
-              "Instruction must be in the same anchor scope function."); 
-   
-       if (!isAssumedDead(I)) 
-         return true; 
-     } 
-   
-     return false; 
-   } 
-   
- public: 
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAIsDead &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// Determine if \p F might catch asynchronous exceptions. 
-   static bool mayCatchAsynchronousExceptions(const Function &F) { 
-     return F.hasPersonalityFn() && !canSimplifyInvokeNoUnwind(&F); 
-   } 
-   
-   /// Returns true if \p BB is assumed dead. 
-   virtual bool isAssumedDead(const BasicBlock *BB) const = 0; 
-   
-   /// Return if the edge from \p From BB to \p To BB is assumed dead. 
-   /// This is specifically useful in AAReachability. 
-   virtual bool isEdgeDead(const BasicBlock *From, const BasicBlock *To) const { 
-     return false; 
-   } 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAIsDead"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AAIsDead 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
-   
-   friend struct Attributor; 
- }; 
-   
- /// State for dereferenceable attribute 
- struct DerefState : AbstractState { 
-   
-   static DerefState getBestState() { return DerefState(); } 
-   static DerefState getBestState(const DerefState &) { return getBestState(); } 
-   
-   /// Return the worst possible representable state. 
-   static DerefState getWorstState() { 
-     DerefState DS; 
-     DS.indicatePessimisticFixpoint(); 
-     return DS; 
-   } 
-   static DerefState getWorstState(const DerefState &) { 
-     return getWorstState(); 
-   } 
-   
-   /// State representing for dereferenceable bytes. 
-   IncIntegerState<> DerefBytesState; 
-   
-   /// Map representing for accessed memory offsets and sizes. 
-   /// A key is Offset and a value is size. 
-   /// If there is a load/store instruction something like, 
-   ///   p[offset] = v; 
-   /// (offset, sizeof(v)) will be inserted to this map. 
-   /// std::map is used because we want to iterate keys in ascending order. 
-   std::map<int64_t, uint64_t> AccessedBytesMap; 
-   
-   /// Helper function to calculate dereferenceable bytes from current known 
-   /// bytes and accessed bytes. 
-   /// 
-   /// int f(int *A){ 
-   ///    *A = 0; 
-   ///    *(A+2) = 2; 
-   ///    *(A+1) = 1; 
-   ///    *(A+10) = 10; 
-   /// } 
-   /// ``` 
-   /// In that case, AccessedBytesMap is `{0:4, 4:4, 8:4, 40:4}`. 
-   /// AccessedBytesMap is std::map so it is iterated in accending order on 
-   /// key(Offset). So KnownBytes will be updated like this: 
-   /// 
-   /// |Access | KnownBytes 
-   /// |(0, 4)| 0 -> 4 
-   /// |(4, 4)| 4 -> 8 
-   /// |(8, 4)| 8 -> 12 
-   /// |(40, 4) | 12 (break) 
-   void computeKnownDerefBytesFromAccessedMap() { 
-     int64_t KnownBytes = DerefBytesState.getKnown(); 
-     for (auto &Access : AccessedBytesMap) { 
-       if (KnownBytes < Access.first) 
-         break; 
-       KnownBytes = std::max(KnownBytes, Access.first + (int64_t)Access.second); 
-     } 
-   
-     DerefBytesState.takeKnownMaximum(KnownBytes); 
-   } 
-   
-   /// State representing that whether the value is globaly dereferenceable. 
-   BooleanState GlobalState; 
-   
-   /// See AbstractState::isValidState() 
-   bool isValidState() const override { return DerefBytesState.isValidState(); } 
-   
-   /// See AbstractState::isAtFixpoint() 
-   bool isAtFixpoint() const override { 
-     return !isValidState() || 
-            (DerefBytesState.isAtFixpoint() && GlobalState.isAtFixpoint()); 
-   } 
-   
-   /// See AbstractState::indicateOptimisticFixpoint(...) 
-   ChangeStatus indicateOptimisticFixpoint() override { 
-     DerefBytesState.indicateOptimisticFixpoint(); 
-     GlobalState.indicateOptimisticFixpoint(); 
-     return ChangeStatus::UNCHANGED; 
-   } 
-   
-   /// See AbstractState::indicatePessimisticFixpoint(...) 
-   ChangeStatus indicatePessimisticFixpoint() override { 
-     DerefBytesState.indicatePessimisticFixpoint(); 
-     GlobalState.indicatePessimisticFixpoint(); 
-     return ChangeStatus::CHANGED; 
-   } 
-   
-   /// Update known dereferenceable bytes. 
-   void takeKnownDerefBytesMaximum(uint64_t Bytes) { 
-     DerefBytesState.takeKnownMaximum(Bytes); 
-   
-     // Known bytes might increase. 
-     computeKnownDerefBytesFromAccessedMap(); 
-   } 
-   
-   /// Update assumed dereferenceable bytes. 
-   void takeAssumedDerefBytesMinimum(uint64_t Bytes) { 
-     DerefBytesState.takeAssumedMinimum(Bytes); 
-   } 
-   
-   /// Add accessed bytes to the map. 
-   void addAccessedBytes(int64_t Offset, uint64_t Size) { 
-     uint64_t &AccessedBytes = AccessedBytesMap[Offset]; 
-     AccessedBytes = std::max(AccessedBytes, Size); 
-   
-     // Known bytes might increase. 
-     computeKnownDerefBytesFromAccessedMap(); 
-   } 
-   
-   /// Equality for DerefState. 
-   bool operator==(const DerefState &R) const { 
-     return this->DerefBytesState == R.DerefBytesState && 
-            this->GlobalState == R.GlobalState; 
-   } 
-   
-   /// Inequality for DerefState. 
-   bool operator!=(const DerefState &R) const { return !(*this == R); } 
-   
-   /// See IntegerStateBase::operator^= 
-   DerefState operator^=(const DerefState &R) { 
-     DerefBytesState ^= R.DerefBytesState; 
-     GlobalState ^= R.GlobalState; 
-     return *this; 
-   } 
-   
-   /// See IntegerStateBase::operator+= 
-   DerefState operator+=(const DerefState &R) { 
-     DerefBytesState += R.DerefBytesState; 
-     GlobalState += R.GlobalState; 
-     return *this; 
-   } 
-   
-   /// See IntegerStateBase::operator&= 
-   DerefState operator&=(const DerefState &R) { 
-     DerefBytesState &= R.DerefBytesState; 
-     GlobalState &= R.GlobalState; 
-     return *this; 
-   } 
-   
-   /// See IntegerStateBase::operator|= 
-   DerefState operator|=(const DerefState &R) { 
-     DerefBytesState |= R.DerefBytesState; 
-     GlobalState |= R.GlobalState; 
-     return *this; 
-   } 
-   
- protected: 
-   const AANonNull *NonNullAA = nullptr; 
- }; 
-   
- /// An abstract interface for all dereferenceable attribute. 
- struct AADereferenceable 
-     : public IRAttribute<Attribute::Dereferenceable, 
-                          StateWrapper<DerefState, AbstractAttribute>> { 
-   AADereferenceable(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Return true if we assume that the underlying value is nonnull. 
-   bool isAssumedNonNull() const { 
-     return NonNullAA && NonNullAA->isAssumedNonNull(); 
-   } 
-   
-   /// Return true if we know that the underlying value is nonnull. 
-   bool isKnownNonNull() const { 
-     return NonNullAA && NonNullAA->isKnownNonNull(); 
-   } 
-   
-   /// Return true if we assume that underlying value is 
-   /// dereferenceable(_or_null) globally. 
-   bool isAssumedGlobal() const { return GlobalState.getAssumed(); } 
-   
-   /// Return true if we know that underlying value is 
-   /// dereferenceable(_or_null) globally. 
-   bool isKnownGlobal() const { return GlobalState.getKnown(); } 
-   
-   /// Return assumed dereferenceable bytes. 
-   uint32_t getAssumedDereferenceableBytes() const { 
-     return DerefBytesState.getAssumed(); 
-   } 
-   
-   /// Return known dereferenceable bytes. 
-   uint32_t getKnownDereferenceableBytes() const { 
-     return DerefBytesState.getKnown(); 
-   } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AADereferenceable &createForPosition(const IRPosition &IRP, 
-                                               Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AADereferenceable"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AADereferenceable 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- using AAAlignmentStateType = 
-     IncIntegerState<uint64_t, Value::MaximumAlignment, 1>; 
- /// An abstract interface for all align attributes. 
- struct AAAlign : public IRAttribute< 
-                      Attribute::Alignment, 
-                      StateWrapper<AAAlignmentStateType, AbstractAttribute>> { 
-   AAAlign(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Return assumed alignment. 
-   Align getAssumedAlign() const { return Align(getAssumed()); } 
-   
-   /// Return known alignment. 
-   Align getKnownAlign() const { return Align(getKnown()); } 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAAlign"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AAAlign 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAAlign &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract interface to track if a value leaves it's defining function 
- /// instance. 
- /// TODO: We should make it a ternary AA tracking uniqueness, and uniqueness 
- /// wrt. the Attributor analysis separately. 
- struct AAInstanceInfo : public StateWrapper<BooleanState, AbstractAttribute> { 
-   AAInstanceInfo(const IRPosition &IRP, Attributor &A) 
-       : StateWrapper<BooleanState, AbstractAttribute>(IRP) {} 
-   
-   /// Return true if we know that the underlying value is unique in its scope 
-   /// wrt. the Attributor analysis. That means it might not be unique but we can 
-   /// still use pointer equality without risking to represent two instances with 
-   /// one `llvm::Value`. 
-   bool isKnownUniqueForAnalysis() const { return isKnown(); } 
-   
-   /// Return true if we assume that the underlying value is unique in its scope 
-   /// wrt. the Attributor analysis. That means it might not be unique but we can 
-   /// still use pointer equality without risking to represent two instances with 
-   /// one `llvm::Value`. 
-   bool isAssumedUniqueForAnalysis() const { return isAssumed(); } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAInstanceInfo &createForPosition(const IRPosition &IRP, 
-                                            Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAInstanceInfo"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAInstanceInfo 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract interface for all nocapture attributes. 
- struct AANoCapture 
-     : public IRAttribute< 
-           Attribute::NoCapture, 
-           StateWrapper<BitIntegerState<uint16_t, 7, 0>, AbstractAttribute>> { 
-   AANoCapture(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// State encoding bits. A set bit in the state means the property holds. 
-   /// NO_CAPTURE is the best possible state, 0 the worst possible state. 
-   enum { 
-     NOT_CAPTURED_IN_MEM = 1 << 0, 
-     NOT_CAPTURED_IN_INT = 1 << 1, 
-     NOT_CAPTURED_IN_RET = 1 << 2, 
-   
-     /// If we do not capture the value in memory or through integers we can only 
-     /// communicate it back as a derived pointer. 
-     NO_CAPTURE_MAYBE_RETURNED = NOT_CAPTURED_IN_MEM | NOT_CAPTURED_IN_INT, 
-   
-     /// If we do not capture the value in memory, through integers, or as a 
-     /// derived pointer we know it is not captured. 
-     NO_CAPTURE = 
-         NOT_CAPTURED_IN_MEM | NOT_CAPTURED_IN_INT | NOT_CAPTURED_IN_RET, 
-   }; 
-   
-   /// Return true if we know that the underlying value is not captured in its 
-   /// respective scope. 
-   bool isKnownNoCapture() const { return isKnown(NO_CAPTURE); } 
-   
-   /// Return true if we assume that the underlying value is not captured in its 
-   /// respective scope. 
-   bool isAssumedNoCapture() const { return isAssumed(NO_CAPTURE); } 
-   
-   /// Return true if we know that the underlying value is not captured in its 
-   /// respective scope but we allow it to escape through a "return". 
-   bool isKnownNoCaptureMaybeReturned() const { 
-     return isKnown(NO_CAPTURE_MAYBE_RETURNED); 
-   } 
-   
-   /// Return true if we assume that the underlying value is not captured in its 
-   /// respective scope but we allow it to escape through a "return". 
-   bool isAssumedNoCaptureMaybeReturned() const { 
-     return isAssumed(NO_CAPTURE_MAYBE_RETURNED); 
-   } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AANoCapture &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AANoCapture"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AANoCapture 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- struct ValueSimplifyStateType : public AbstractState { 
-   
-   ValueSimplifyStateType(Type *Ty) : Ty(Ty) {} 
-   
-   static ValueSimplifyStateType getBestState(Type *Ty) { 
-     return ValueSimplifyStateType(Ty); 
-   } 
-   static ValueSimplifyStateType getBestState(const ValueSimplifyStateType &VS) { 
-     return getBestState(VS.Ty); 
-   } 
-   
-   /// Return the worst possible representable state. 
-   static ValueSimplifyStateType getWorstState(Type *Ty) { 
-     ValueSimplifyStateType DS(Ty); 
-     DS.indicatePessimisticFixpoint(); 
-     return DS; 
-   } 
-   static ValueSimplifyStateType 
-   getWorstState(const ValueSimplifyStateType &VS) { 
-     return getWorstState(VS.Ty); 
-   } 
-   
-   /// See AbstractState::isValidState(...) 
-   bool isValidState() const override { return BS.isValidState(); } 
-   
-   /// See AbstractState::isAtFixpoint(...) 
-   bool isAtFixpoint() const override { return BS.isAtFixpoint(); } 
-   
-   /// Return the assumed state encoding. 
-   ValueSimplifyStateType getAssumed() { return *this; } 
-   const ValueSimplifyStateType &getAssumed() const { return *this; } 
-   
-   /// See AbstractState::indicatePessimisticFixpoint(...) 
-   ChangeStatus indicatePessimisticFixpoint() override { 
-     return BS.indicatePessimisticFixpoint(); 
-   } 
-   
-   /// See AbstractState::indicateOptimisticFixpoint(...) 
-   ChangeStatus indicateOptimisticFixpoint() override { 
-     return BS.indicateOptimisticFixpoint(); 
-   } 
-   
-   /// "Clamp" this state with \p PVS. 
-   ValueSimplifyStateType operator^=(const ValueSimplifyStateType &VS) { 
-     BS ^= VS.BS; 
-     unionAssumed(VS.SimplifiedAssociatedValue); 
-     return *this; 
-   } 
-   
-   bool operator==(const ValueSimplifyStateType &RHS) const { 
-     if (isValidState() != RHS.isValidState()) 
-       return false; 
-     if (!isValidState() && !RHS.isValidState()) 
-       return true; 
-     return SimplifiedAssociatedValue == RHS.SimplifiedAssociatedValue; 
-   } 
-   
- protected: 
-   /// The type of the original value. 
-   Type *Ty; 
-   
-   /// Merge \p Other into the currently assumed simplified value 
-   bool unionAssumed(std::optional<Value *> Other); 
-   
-   /// Helper to track validity and fixpoint 
-   BooleanState BS; 
-   
-   /// An assumed simplified value. Initially, it is set to std::nullopt, which 
-   /// means that the value is not clear under current assumption. If in the 
-   /// pessimistic state, getAssumedSimplifiedValue doesn't return this value but 
-   /// returns orignal associated value. 
-   std::optional<Value *> SimplifiedAssociatedValue; 
- }; 
-   
- /// An abstract interface for value simplify abstract attribute. 
- struct AAValueSimplify 
-     : public StateWrapper<ValueSimplifyStateType, AbstractAttribute, Type *> { 
-   using Base = StateWrapper<ValueSimplifyStateType, AbstractAttribute, Type *>; 
-   AAValueSimplify(const IRPosition &IRP, Attributor &A) 
-       : Base(IRP, IRP.getAssociatedType()) {} 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAValueSimplify &createForPosition(const IRPosition &IRP, 
-                                             Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAValueSimplify"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAValueSimplify 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
-   
- private: 
-   /// Return an assumed simplified value if a single candidate is found. If 
-   /// there cannot be one, return original value. If it is not clear yet, return 
-   /// std::nullopt. 
-   /// 
-   /// Use `Attributor::getAssumedSimplified` for value simplification. 
-   virtual std::optional<Value *> 
-   getAssumedSimplifiedValue(Attributor &A) const = 0; 
-   
-   friend struct Attributor; 
- }; 
-   
- struct AAHeapToStack : public StateWrapper<BooleanState, AbstractAttribute> { 
-   using Base = StateWrapper<BooleanState, AbstractAttribute>; 
-   AAHeapToStack(const IRPosition &IRP, Attributor &A) : Base(IRP) {} 
-   
-   /// Returns true if HeapToStack conversion is assumed to be possible. 
-   virtual bool isAssumedHeapToStack(const CallBase &CB) const = 0; 
-   
-   /// Returns true if HeapToStack conversion is assumed and the CB is a 
-   /// callsite to a free operation to be removed. 
-   virtual bool isAssumedHeapToStackRemovedFree(CallBase &CB) const = 0; 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAHeapToStack &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAHeapToStack"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AAHeapToStack 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract interface for privatizability. 
- /// 
- /// A pointer is privatizable if it can be replaced by a new, private one. 
- /// Privatizing pointer reduces the use count, interaction between unrelated 
- /// code parts. 
- /// 
- /// In order for a pointer to be privatizable its value cannot be observed 
- /// (=nocapture), it is (for now) not written (=readonly & noalias), we know 
- /// what values are necessary to make the private copy look like the original 
- /// one, and the values we need can be loaded (=dereferenceable). 
- struct AAPrivatizablePtr 
-     : public StateWrapper<BooleanState, AbstractAttribute> { 
-   using Base = StateWrapper<BooleanState, AbstractAttribute>; 
-   AAPrivatizablePtr(const IRPosition &IRP, Attributor &A) : Base(IRP) {} 
-   
-   /// Returns true if pointer privatization is assumed to be possible. 
-   bool isAssumedPrivatizablePtr() const { return getAssumed(); } 
-   
-   /// Returns true if pointer privatization is known to be possible. 
-   bool isKnownPrivatizablePtr() const { return getKnown(); } 
-   
-   /// Return the type we can choose for a private copy of the underlying 
-   /// value. std::nullopt means it is not clear yet, nullptr means there is 
-   /// none. 
-   virtual std::optional<Type *> getPrivatizableType() const = 0; 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAPrivatizablePtr &createForPosition(const IRPosition &IRP, 
-                                               Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAPrivatizablePtr"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAPricatizablePtr 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract interface for memory access kind related attributes 
- /// (readnone/readonly/writeonly). 
- struct AAMemoryBehavior 
-     : public IRAttribute< 
-           Attribute::ReadNone, 
-           StateWrapper<BitIntegerState<uint8_t, 3>, AbstractAttribute>> { 
-   AAMemoryBehavior(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// State encoding bits. A set bit in the state means the property holds. 
-   /// BEST_STATE is the best possible state, 0 the worst possible state. 
-   enum { 
-     NO_READS = 1 << 0, 
-     NO_WRITES = 1 << 1, 
-     NO_ACCESSES = NO_READS | NO_WRITES, 
-   
-     BEST_STATE = NO_ACCESSES, 
-   }; 
-   static_assert(BEST_STATE == getBestState(), "Unexpected BEST_STATE value"); 
-   
-   /// Return true if we know that the underlying value is not read or accessed 
-   /// in its respective scope. 
-   bool isKnownReadNone() const { return isKnown(NO_ACCESSES); } 
-   
-   /// Return true if we assume that the underlying value is not read or accessed 
-   /// in its respective scope. 
-   bool isAssumedReadNone() const { return isAssumed(NO_ACCESSES); } 
-   
-   /// Return true if we know that the underlying value is not accessed 
-   /// (=written) in its respective scope. 
-   bool isKnownReadOnly() const { return isKnown(NO_WRITES); } 
-   
-   /// Return true if we assume that the underlying value is not accessed 
-   /// (=written) in its respective scope. 
-   bool isAssumedReadOnly() const { return isAssumed(NO_WRITES); } 
-   
-   /// Return true if we know that the underlying value is not read in its 
-   /// respective scope. 
-   bool isKnownWriteOnly() const { return isKnown(NO_READS); } 
-   
-   /// Return true if we assume that the underlying value is not read in its 
-   /// respective scope. 
-   bool isAssumedWriteOnly() const { return isAssumed(NO_READS); } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAMemoryBehavior &createForPosition(const IRPosition &IRP, 
-                                              Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAMemoryBehavior"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAMemoryBehavior 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract interface for all memory location attributes 
- /// (readnone/argmemonly/inaccessiblememonly/inaccessibleorargmemonly). 
- struct AAMemoryLocation 
-     : public IRAttribute< 
-           Attribute::ReadNone, 
-           StateWrapper<BitIntegerState<uint32_t, 511>, AbstractAttribute>> { 
-   using MemoryLocationsKind = StateType::base_t; 
-   
-   AAMemoryLocation(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Encoding of different locations that could be accessed by a memory 
-   /// access. 
-   enum { 
-     ALL_LOCATIONS = 0, 
-     NO_LOCAL_MEM = 1 << 0, 
-     NO_CONST_MEM = 1 << 1, 
-     NO_GLOBAL_INTERNAL_MEM = 1 << 2, 
-     NO_GLOBAL_EXTERNAL_MEM = 1 << 3, 
-     NO_GLOBAL_MEM = NO_GLOBAL_INTERNAL_MEM | NO_GLOBAL_EXTERNAL_MEM, 
-     NO_ARGUMENT_MEM = 1 << 4, 
-     NO_INACCESSIBLE_MEM = 1 << 5, 
-     NO_MALLOCED_MEM = 1 << 6, 
-     NO_UNKOWN_MEM = 1 << 7, 
-     NO_LOCATIONS = NO_LOCAL_MEM | NO_CONST_MEM | NO_GLOBAL_INTERNAL_MEM | 
-                    NO_GLOBAL_EXTERNAL_MEM | NO_ARGUMENT_MEM | 
-                    NO_INACCESSIBLE_MEM | NO_MALLOCED_MEM | NO_UNKOWN_MEM, 
-   
-     // Helper bit to track if we gave up or not. 
-     VALID_STATE = NO_LOCATIONS + 1, 
-   
-     BEST_STATE = NO_LOCATIONS | VALID_STATE, 
-   }; 
-   static_assert(BEST_STATE == getBestState(), "Unexpected BEST_STATE value"); 
-   
-   /// Return true if we know that the associated functions has no observable 
-   /// accesses. 
-   bool isKnownReadNone() const { return isKnown(NO_LOCATIONS); } 
-   
-   /// Return true if we assume that the associated functions has no observable 
-   /// accesses. 
-   bool isAssumedReadNone() const { 
-     return isAssumed(NO_LOCATIONS) || isAssumedStackOnly(); 
-   } 
-   
-   /// Return true if we know that the associated functions has at most 
-   /// local/stack accesses. 
-   bool isKnowStackOnly() const { 
-     return isKnown(inverseLocation(NO_LOCAL_MEM, true, true)); 
-   } 
-   
-   /// Return true if we assume that the associated functions has at most 
-   /// local/stack accesses. 
-   bool isAssumedStackOnly() const { 
-     return isAssumed(inverseLocation(NO_LOCAL_MEM, true, true)); 
-   } 
-   
-   /// Return true if we know that the underlying value will only access 
-   /// inaccesible memory only (see Attribute::InaccessibleMemOnly). 
-   bool isKnownInaccessibleMemOnly() const { 
-     return isKnown(inverseLocation(NO_INACCESSIBLE_MEM, true, true)); 
-   } 
-   
-   /// Return true if we assume that the underlying value will only access 
-   /// inaccesible memory only (see Attribute::InaccessibleMemOnly). 
-   bool isAssumedInaccessibleMemOnly() const { 
-     return isAssumed(inverseLocation(NO_INACCESSIBLE_MEM, true, true)); 
-   } 
-   
-   /// Return true if we know that the underlying value will only access 
-   /// argument pointees (see Attribute::ArgMemOnly). 
-   bool isKnownArgMemOnly() const { 
-     return isKnown(inverseLocation(NO_ARGUMENT_MEM, true, true)); 
-   } 
-   
-   /// Return true if we assume that the underlying value will only access 
-   /// argument pointees (see Attribute::ArgMemOnly). 
-   bool isAssumedArgMemOnly() const { 
-     return isAssumed(inverseLocation(NO_ARGUMENT_MEM, true, true)); 
-   } 
-   
-   /// Return true if we know that the underlying value will only access 
-   /// inaccesible memory or argument pointees (see 
-   /// Attribute::InaccessibleOrArgMemOnly). 
-   bool isKnownInaccessibleOrArgMemOnly() const { 
-     return isKnown( 
-         inverseLocation(NO_INACCESSIBLE_MEM | NO_ARGUMENT_MEM, true, true)); 
-   } 
-   
-   /// Return true if we assume that the underlying value will only access 
-   /// inaccesible memory or argument pointees (see 
-   /// Attribute::InaccessibleOrArgMemOnly). 
-   bool isAssumedInaccessibleOrArgMemOnly() const { 
-     return isAssumed( 
-         inverseLocation(NO_INACCESSIBLE_MEM | NO_ARGUMENT_MEM, true, true)); 
-   } 
-   
-   /// Return true if the underlying value may access memory through arguement 
-   /// pointers of the associated function, if any. 
-   bool mayAccessArgMem() const { return !isAssumed(NO_ARGUMENT_MEM); } 
-   
-   /// Return true if only the memory locations specififed by \p MLK are assumed 
-   /// to be accessed by the associated function. 
-   bool isAssumedSpecifiedMemOnly(MemoryLocationsKind MLK) const { 
-     return isAssumed(MLK); 
-   } 
-   
-   /// Return the locations that are assumed to be not accessed by the associated 
-   /// function, if any. 
-   MemoryLocationsKind getAssumedNotAccessedLocation() const { 
-     return getAssumed(); 
-   } 
-   
-   /// Return the inverse of location \p Loc, thus for NO_XXX the return 
-   /// describes ONLY_XXX. The flags \p AndLocalMem and \p AndConstMem determine 
-   /// if local (=stack) and constant memory are allowed as well. Most of the 
-   /// time we do want them to be included, e.g., argmemonly allows accesses via 
-   /// argument pointers or local or constant memory accesses. 
-   static MemoryLocationsKind 
-   inverseLocation(MemoryLocationsKind Loc, bool AndLocalMem, bool AndConstMem) { 
-     return NO_LOCATIONS & ~(Loc | (AndLocalMem ? NO_LOCAL_MEM : 0) | 
-                             (AndConstMem ? NO_CONST_MEM : 0)); 
-   }; 
-   
-   /// Return the locations encoded by \p MLK as a readable string. 
-   static std::string getMemoryLocationsAsStr(MemoryLocationsKind MLK); 
-   
-   /// Simple enum to distinguish read/write/read-write accesses. 
-   enum AccessKind { 
-     NONE = 0, 
-     READ = 1 << 0, 
-     WRITE = 1 << 1, 
-     READ_WRITE = READ | WRITE, 
-   }; 
-   
-   /// Check \p Pred on all accesses to the memory kinds specified by \p MLK. 
-   /// 
-   /// This method will evaluate \p Pred on all accesses (access instruction + 
-   /// underlying accessed memory pointer) and it will return true if \p Pred 
-   /// holds every time. 
-   virtual bool checkForAllAccessesToMemoryKind( 
-       function_ref<bool(const Instruction *, const Value *, AccessKind, 
-                         MemoryLocationsKind)> 
-           Pred, 
-       MemoryLocationsKind MLK) const = 0; 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAMemoryLocation &createForPosition(const IRPosition &IRP, 
-                                              Attributor &A); 
-   
-   /// See AbstractState::getAsStr(). 
-   const std::string getAsStr() const override { 
-     return getMemoryLocationsAsStr(getAssumedNotAccessedLocation()); 
-   } 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAMemoryLocation"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAMemoryLocation 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract interface for range value analysis. 
- struct AAValueConstantRange 
-     : public StateWrapper<IntegerRangeState, AbstractAttribute, uint32_t> { 
-   using Base = StateWrapper<IntegerRangeState, AbstractAttribute, uint32_t>; 
-   AAValueConstantRange(const IRPosition &IRP, Attributor &A) 
-       : Base(IRP, IRP.getAssociatedType()->getIntegerBitWidth()) {} 
-   
-   /// See AbstractAttribute::getState(...). 
-   IntegerRangeState &getState() override { return *this; } 
-   const IntegerRangeState &getState() const override { return *this; } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAValueConstantRange &createForPosition(const IRPosition &IRP, 
-                                                  Attributor &A); 
-   
-   /// Return an assumed range for the associated value a program point \p CtxI. 
-   /// If \p I is nullptr, simply return an assumed range. 
-   virtual ConstantRange 
-   getAssumedConstantRange(Attributor &A, 
-                           const Instruction *CtxI = nullptr) const = 0; 
-   
-   /// Return a known range for the associated value at a program point \p CtxI. 
-   /// If \p I is nullptr, simply return a known range. 
-   virtual ConstantRange 
-   getKnownConstantRange(Attributor &A, 
-                         const Instruction *CtxI = nullptr) const = 0; 
-   
-   /// Return an assumed constant for the associated value a program point \p 
-   /// CtxI. 
-   std::optional<Constant *> 
-   getAssumedConstant(Attributor &A, const Instruction *CtxI = nullptr) const { 
-     ConstantRange RangeV = getAssumedConstantRange(A, CtxI); 
-     if (auto *C = RangeV.getSingleElement()) { 
-       Type *Ty = getAssociatedValue().getType(); 
-       return cast_or_null<Constant>( 
-           AA::getWithType(*ConstantInt::get(Ty->getContext(), *C), *Ty)); 
-     } 
-     if (RangeV.isEmptySet()) 
-       return std::nullopt; 
-     return nullptr; 
-   } 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAValueConstantRange"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAValueConstantRange 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// A class for a set state. 
- /// The assumed boolean state indicates whether the corresponding set is full 
- /// set or not. If the assumed state is false, this is the worst state. The 
- /// worst state (invalid state) of set of potential values is when the set 
- /// contains every possible value (i.e. we cannot in any way limit the value 
- /// that the target position can take). That never happens naturally, we only 
- /// force it. As for the conditions under which we force it, see 
- /// AAPotentialConstantValues. 
- template <typename MemberTy> struct PotentialValuesState : AbstractState { 
-   using SetTy = SmallSetVector<MemberTy, 8>; 
-   
-   PotentialValuesState() : IsValidState(true), UndefIsContained(false) {} 
-   
-   PotentialValuesState(bool IsValid) 
-       : IsValidState(IsValid), UndefIsContained(false) {} 
-   
-   /// See AbstractState::isValidState(...) 
-   bool isValidState() const override { return IsValidState.isValidState(); } 
-   
-   /// See AbstractState::isAtFixpoint(...) 
-   bool isAtFixpoint() const override { return IsValidState.isAtFixpoint(); } 
-   
-   /// See AbstractState::indicatePessimisticFixpoint(...) 
-   ChangeStatus indicatePessimisticFixpoint() override { 
-     return IsValidState.indicatePessimisticFixpoint(); 
-   } 
-   
-   /// See AbstractState::indicateOptimisticFixpoint(...) 
-   ChangeStatus indicateOptimisticFixpoint() override { 
-     return IsValidState.indicateOptimisticFixpoint(); 
-   } 
-   
-   /// Return the assumed state 
-   PotentialValuesState &getAssumed() { return *this; } 
-   const PotentialValuesState &getAssumed() const { return *this; } 
-   
-   /// Return this set. We should check whether this set is valid or not by 
-   /// isValidState() before calling this function. 
-   const SetTy &getAssumedSet() const { 
-     assert(isValidState() && "This set shoud not be used when it is invalid!"); 
-     return Set; 
-   } 
-   
-   /// Returns whether this state contains an undef value or not. 
-   bool undefIsContained() const { 
-     assert(isValidState() && "This flag shoud not be used when it is invalid!"); 
-     return UndefIsContained; 
-   } 
-   
-   bool operator==(const PotentialValuesState &RHS) const { 
-     if (isValidState() != RHS.isValidState()) 
-       return false; 
-     if (!isValidState() && !RHS.isValidState()) 
-       return true; 
-     if (undefIsContained() != RHS.undefIsContained()) 
-       return false; 
-     return Set == RHS.getAssumedSet(); 
-   } 
-   
-   /// Maximum number of potential values to be tracked. 
-   /// This is set by -attributor-max-potential-values command line option 
-   static unsigned MaxPotentialValues; 
-   
-   /// Return empty set as the best state of potential values. 
-   static PotentialValuesState getBestState() { 
-     return PotentialValuesState(true); 
-   } 
-   
-   static PotentialValuesState getBestState(const PotentialValuesState &PVS) { 
-     return getBestState(); 
-   } 
-   
-   /// Return full set as the worst state of potential values. 
-   static PotentialValuesState getWorstState() { 
-     return PotentialValuesState(false); 
-   } 
-   
-   /// Union assumed set with the passed value. 
-   void unionAssumed(const MemberTy &C) { insert(C); } 
-   
-   /// Union assumed set with assumed set of the passed state \p PVS. 
-   void unionAssumed(const PotentialValuesState &PVS) { unionWith(PVS); } 
-   
-   /// Union assumed set with an undef value. 
-   void unionAssumedWithUndef() { unionWithUndef(); } 
-   
-   /// "Clamp" this state with \p PVS. 
-   PotentialValuesState operator^=(const PotentialValuesState &PVS) { 
-     IsValidState ^= PVS.IsValidState; 
-     unionAssumed(PVS); 
-     return *this; 
-   } 
-   
-   PotentialValuesState operator&=(const PotentialValuesState &PVS) { 
-     IsValidState &= PVS.IsValidState; 
-     unionAssumed(PVS); 
-     return *this; 
-   } 
-   
-   bool contains(const MemberTy &V) const { 
-     return !isValidState() ? true : Set.contains(V); 
-   } 
-   
- protected: 
-   SetTy &getAssumedSet() { 
-     assert(isValidState() && "This set shoud not be used when it is invalid!"); 
-     return Set; 
-   } 
-   
- private: 
-   /// Check the size of this set, and invalidate when the size is no 
-   /// less than \p MaxPotentialValues threshold. 
-   void checkAndInvalidate() { 
-     if (Set.size() >= MaxPotentialValues) 
-       indicatePessimisticFixpoint(); 
-     else 
-       reduceUndefValue(); 
-   } 
-   
-   /// If this state contains both undef and not undef, we can reduce 
-   /// undef to the not undef value. 
-   void reduceUndefValue() { UndefIsContained = UndefIsContained & Set.empty(); } 
-   
-   /// Insert an element into this set. 
-   void insert(const MemberTy &C) { 
-     if (!isValidState()) 
-       return; 
-     Set.insert(C); 
-     checkAndInvalidate(); 
-   } 
-   
-   /// Take union with R. 
-   void unionWith(const PotentialValuesState &R) { 
-     /// If this is a full set, do nothing. 
-     if (!isValidState()) 
-       return; 
-     /// If R is full set, change L to a full set. 
-     if (!R.isValidState()) { 
-       indicatePessimisticFixpoint(); 
-       return; 
-     } 
-     for (const MemberTy &C : R.Set) 
-       Set.insert(C); 
-     UndefIsContained |= R.undefIsContained(); 
-     checkAndInvalidate(); 
-   } 
-   
-   /// Take union with an undef value. 
-   void unionWithUndef() { 
-     UndefIsContained = true; 
-     reduceUndefValue(); 
-   } 
-   
-   /// Take intersection with R. 
-   void intersectWith(const PotentialValuesState &R) { 
-     /// If R is a full set, do nothing. 
-     if (!R.isValidState()) 
-       return; 
-     /// If this is a full set, change this to R. 
-     if (!isValidState()) { 
-       *this = R; 
-       return; 
-     } 
-     SetTy IntersectSet; 
-     for (const MemberTy &C : Set) { 
-       if (R.Set.count(C)) 
-         IntersectSet.insert(C); 
-     } 
-     Set = IntersectSet; 
-     UndefIsContained &= R.undefIsContained(); 
-     reduceUndefValue(); 
-   } 
-   
-   /// A helper state which indicate whether this state is valid or not. 
-   BooleanState IsValidState; 
-   
-   /// Container for potential values 
-   SetTy Set; 
-   
-   /// Flag for undef value 
-   bool UndefIsContained; 
- }; 
-   
- using PotentialConstantIntValuesState = PotentialValuesState<APInt>; 
- using PotentialLLVMValuesState = 
-     PotentialValuesState<std::pair<AA::ValueAndContext, AA::ValueScope>>; 
-   
- raw_ostream &operator<<(raw_ostream &OS, 
-                         const PotentialConstantIntValuesState &R); 
- raw_ostream &operator<<(raw_ostream &OS, const PotentialLLVMValuesState &R); 
-   
- /// An abstract interface for potential values analysis. 
- /// 
- /// This AA collects potential values for each IR position. 
- /// An assumed set of potential values is initialized with the empty set (the 
- /// best state) and it will grow monotonically as we find more potential values 
- /// for this position. 
- /// The set might be forced to the worst state, that is, to contain every 
- /// possible value for this position in 2 cases. 
- ///   1. We surpassed the \p MaxPotentialValues threshold. This includes the 
- ///      case that this position is affected (e.g. because of an operation) by a 
- ///      Value that is in the worst state. 
- ///   2. We tried to initialize on a Value that we cannot handle (e.g. an 
- ///      operator we do not currently handle). 
- /// 
- /// For non constant integers see AAPotentialValues. 
- struct AAPotentialConstantValues 
-     : public StateWrapper<PotentialConstantIntValuesState, AbstractAttribute> { 
-   using Base = StateWrapper<PotentialConstantIntValuesState, AbstractAttribute>; 
-   AAPotentialConstantValues(const IRPosition &IRP, Attributor &A) : Base(IRP) {} 
-   
-   /// See AbstractAttribute::getState(...). 
-   PotentialConstantIntValuesState &getState() override { return *this; } 
-   const PotentialConstantIntValuesState &getState() const override { 
-     return *this; 
-   } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAPotentialConstantValues &createForPosition(const IRPosition &IRP, 
-                                                       Attributor &A); 
-   
-   /// Return assumed constant for the associated value 
-   std::optional<Constant *> 
-   getAssumedConstant(Attributor &A, const Instruction *CtxI = nullptr) const { 
-     if (!isValidState()) 
-       return nullptr; 
-     if (getAssumedSet().size() == 1) { 
-       Type *Ty = getAssociatedValue().getType(); 
-       return cast_or_null<Constant>(AA::getWithType( 
-           *ConstantInt::get(Ty->getContext(), *(getAssumedSet().begin())), 
-           *Ty)); 
-     } 
-     if (getAssumedSet().size() == 0) { 
-       if (undefIsContained()) 
-         return UndefValue::get(getAssociatedValue().getType()); 
-       return std::nullopt; 
-     } 
-   
-     return nullptr; 
-   } 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { 
-     return "AAPotentialConstantValues"; 
-   } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAPotentialConstantValues 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- struct AAPotentialValues 
-     : public StateWrapper<PotentialLLVMValuesState, AbstractAttribute> { 
-   using Base = StateWrapper<PotentialLLVMValuesState, AbstractAttribute>; 
-   AAPotentialValues(const IRPosition &IRP, Attributor &A) : Base(IRP) {} 
-   
-   /// See AbstractAttribute::getState(...). 
-   PotentialLLVMValuesState &getState() override { return *this; } 
-   const PotentialLLVMValuesState &getState() const override { return *this; } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAPotentialValues &createForPosition(const IRPosition &IRP, 
-                                               Attributor &A); 
-   
-   /// Extract the single value in \p Values if any. 
-   static Value *getSingleValue(Attributor &A, const AbstractAttribute &AA, 
-                                const IRPosition &IRP, 
-                                SmallVectorImpl<AA::ValueAndContext> &Values); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAPotentialValues"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAPotentialValues 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
-   
- private: 
-   virtual bool 
-   getAssumedSimplifiedValues(Attributor &A, 
-                              SmallVectorImpl<AA::ValueAndContext> &Values, 
-                              AA::ValueScope) const = 0; 
-   
-   friend struct Attributor; 
- }; 
-   
- /// An abstract interface for all noundef attributes. 
- struct AANoUndef 
-     : public IRAttribute<Attribute::NoUndef, 
-                          StateWrapper<BooleanState, AbstractAttribute>> { 
-   AANoUndef(const IRPosition &IRP, Attributor &A) : IRAttribute(IRP) {} 
-   
-   /// Return true if we assume that the underlying value is noundef. 
-   bool isAssumedNoUndef() const { return getAssumed(); } 
-   
-   /// Return true if we know that underlying value is noundef. 
-   bool isKnownNoUndef() const { return getKnown(); } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AANoUndef &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AANoUndef"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AANoUndef 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- struct AACallGraphNode; 
- struct AACallEdges; 
-   
- /// An Iterator for call edges, creates AACallEdges attributes in a lazy way. 
- /// This iterator becomes invalid if the underlying edge list changes. 
- /// So This shouldn't outlive a iteration of Attributor. 
- class AACallEdgeIterator 
-     : public iterator_adaptor_base<AACallEdgeIterator, 
-                                    SetVector<Function *>::iterator> { 
-   AACallEdgeIterator(Attributor &A, SetVector<Function *>::iterator Begin) 
-       : iterator_adaptor_base(Begin), A(A) {} 
-   
- public: 
-   AACallGraphNode *operator*() const; 
-   
- private: 
-   Attributor &A; 
-   friend AACallEdges; 
-   friend AttributorCallGraph; 
- }; 
-   
- struct AACallGraphNode { 
-   AACallGraphNode(Attributor &A) : A(A) {} 
-   virtual ~AACallGraphNode() = default; 
-   
-   virtual AACallEdgeIterator optimisticEdgesBegin() const = 0; 
-   virtual AACallEdgeIterator optimisticEdgesEnd() const = 0; 
-   
-   /// Iterator range for exploring the call graph. 
-   iterator_range<AACallEdgeIterator> optimisticEdgesRange() const { 
-     return iterator_range<AACallEdgeIterator>(optimisticEdgesBegin(), 
-                                               optimisticEdgesEnd()); 
-   } 
-   
- protected: 
-   /// Reference to Attributor needed for GraphTraits implementation. 
-   Attributor &A; 
- }; 
-   
- /// An abstract state for querying live call edges. 
- /// This interface uses the Attributor's optimistic liveness 
- /// information to compute the edges that are alive. 
- struct AACallEdges : public StateWrapper<BooleanState, AbstractAttribute>, 
-                      AACallGraphNode { 
-   using Base = StateWrapper<BooleanState, AbstractAttribute>; 
-   
-   AACallEdges(const IRPosition &IRP, Attributor &A) 
-       : Base(IRP), AACallGraphNode(A) {} 
-   
-   /// Get the optimistic edges. 
-   virtual const SetVector<Function *> &getOptimisticEdges() const = 0; 
-   
-   /// Is there any call with a unknown callee. 
-   virtual bool hasUnknownCallee() const = 0; 
-   
-   /// Is there any call with a unknown callee, excluding any inline asm. 
-   virtual bool hasNonAsmUnknownCallee() const = 0; 
-   
-   /// Iterator for exploring the call graph. 
-   AACallEdgeIterator optimisticEdgesBegin() const override { 
-     return AACallEdgeIterator(A, getOptimisticEdges().begin()); 
-   } 
-   
-   /// Iterator for exploring the call graph. 
-   AACallEdgeIterator optimisticEdgesEnd() const override { 
-     return AACallEdgeIterator(A, getOptimisticEdges().end()); 
-   } 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AACallEdges &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AACallEdges"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AACallEdges. 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- // Synthetic root node for the Attributor's internal call graph. 
- struct AttributorCallGraph : public AACallGraphNode { 
-   AttributorCallGraph(Attributor &A) : AACallGraphNode(A) {} 
-   virtual ~AttributorCallGraph() = default; 
-   
-   AACallEdgeIterator optimisticEdgesBegin() const override { 
-     return AACallEdgeIterator(A, A.Functions.begin()); 
-   } 
-   
-   AACallEdgeIterator optimisticEdgesEnd() const override { 
-     return AACallEdgeIterator(A, A.Functions.end()); 
-   } 
-   
-   /// Force populate the entire call graph. 
-   void populateAll() const { 
-     for (const AACallGraphNode *AA : optimisticEdgesRange()) { 
-       // Nothing else to do here. 
-       (void)AA; 
-     } 
-   } 
-   
-   void print(); 
- }; 
-   
- template <> struct GraphTraits<AACallGraphNode *> { 
-   using NodeRef = AACallGraphNode *; 
-   using ChildIteratorType = AACallEdgeIterator; 
-   
-   static AACallEdgeIterator child_begin(AACallGraphNode *Node) { 
-     return Node->optimisticEdgesBegin(); 
-   } 
-   
-   static AACallEdgeIterator child_end(AACallGraphNode *Node) { 
-     return Node->optimisticEdgesEnd(); 
-   } 
- }; 
-   
- template <> 
- struct GraphTraits<AttributorCallGraph *> 
-     : public GraphTraits<AACallGraphNode *> { 
-   using nodes_iterator = AACallEdgeIterator; 
-   
-   static AACallGraphNode *getEntryNode(AttributorCallGraph *G) { 
-     return static_cast<AACallGraphNode *>(G); 
-   } 
-   
-   static AACallEdgeIterator nodes_begin(const AttributorCallGraph *G) { 
-     return G->optimisticEdgesBegin(); 
-   } 
-   
-   static AACallEdgeIterator nodes_end(const AttributorCallGraph *G) { 
-     return G->optimisticEdgesEnd(); 
-   } 
- }; 
-   
- template <> 
- struct DOTGraphTraits<AttributorCallGraph *> : public DefaultDOTGraphTraits { 
-   DOTGraphTraits(bool Simple = false) : DefaultDOTGraphTraits(Simple) {} 
-   
-   std::string getNodeLabel(const AACallGraphNode *Node, 
-                            const AttributorCallGraph *Graph) { 
-     const AACallEdges *AACE = static_cast<const AACallEdges *>(Node); 
-     return AACE->getAssociatedFunction()->getName().str(); 
-   } 
-   
-   static bool isNodeHidden(const AACallGraphNode *Node, 
-                            const AttributorCallGraph *Graph) { 
-     // Hide the synth root. 
-     return static_cast<const AACallGraphNode *>(Graph) == Node; 
-   } 
- }; 
-   
- struct AAExecutionDomain 
-     : public StateWrapper<BooleanState, AbstractAttribute> { 
-   using Base = StateWrapper<BooleanState, AbstractAttribute>; 
-   AAExecutionDomain(const IRPosition &IRP, Attributor &A) : Base(IRP) {} 
-   
-   /// Summary about the execution domain of a block or instruction. 
-   struct ExecutionDomainTy { 
-     using BarriersSetTy = SmallPtrSet<CallBase *, 2>; 
-     using AssumesSetTy = SmallPtrSet<AssumeInst *, 4>; 
-   
-     void addAssumeInst(Attributor &A, AssumeInst &AI) { 
-       EncounteredAssumes.insert(&AI); 
-     } 
-   
-     void addAlignedBarrier(Attributor &A, CallBase &CB) { 
-       AlignedBarriers.insert(&CB); 
-     } 
-   
-     void clearAssumeInstAndAlignedBarriers() { 
-       EncounteredAssumes.clear(); 
-       AlignedBarriers.clear(); 
-     } 
-   
-     bool IsExecutedByInitialThreadOnly = true; 
-     bool IsReachedFromAlignedBarrierOnly = true; 
-     bool IsReachingAlignedBarrierOnly = true; 
-     bool EncounteredNonLocalSideEffect = false; 
-     BarriersSetTy AlignedBarriers; 
-     AssumesSetTy EncounteredAssumes; 
-   }; 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAExecutionDomain &createForPosition(const IRPosition &IRP, 
-                                               Attributor &A); 
-   
-   /// See AbstractAttribute::getName(). 
-   const std::string getName() const override { return "AAExecutionDomain"; } 
-   
-   /// See AbstractAttribute::getIdAddr(). 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// Check if an instruction is executed only by the initial thread. 
-   bool isExecutedByInitialThreadOnly(const Instruction &I) const { 
-     return isExecutedByInitialThreadOnly(*I.getParent()); 
-   } 
-   
-   /// Check if a basic block is executed only by the initial thread. 
-   virtual bool isExecutedByInitialThreadOnly(const BasicBlock &) const = 0; 
-   
-   /// Check if the instruction \p I is executed in an aligned region, that is, 
-   /// the synchronizing effects before and after \p I are both aligned barriers. 
-   /// This effectively means all threads execute \p I together. 
-   virtual bool isExecutedInAlignedRegion(Attributor &A, 
-                                          const Instruction &I) const = 0; 
-   
-   virtual ExecutionDomainTy getExecutionDomain(const BasicBlock &) const = 0; 
-   virtual ExecutionDomainTy getExecutionDomain(const CallBase &) const = 0; 
-   virtual ExecutionDomainTy getFunctionExecutionDomain() const = 0; 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAExecutionDomain. 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract Attribute for computing reachability between functions. 
- struct AAInterFnReachability 
-     : public StateWrapper<BooleanState, AbstractAttribute> { 
-   using Base = StateWrapper<BooleanState, AbstractAttribute>; 
-   
-   AAInterFnReachability(const IRPosition &IRP, Attributor &A) : Base(IRP) {} 
-   
-   /// If the function represented by this possition can reach \p Fn. 
-   bool canReach(Attributor &A, const Function &Fn) const { 
-     Function *Scope = getAnchorScope(); 
-     if (!Scope || Scope->isDeclaration()) 
-       return true; 
-     return instructionCanReach(A, Scope->getEntryBlock().front(), Fn); 
-   } 
-   
-   /// Can  \p Inst reach \p Fn. 
-   /// See also AA::isPotentiallyReachable. 
-   virtual bool instructionCanReach( 
-       Attributor &A, const Instruction &Inst, const Function &Fn, 
-       const AA::InstExclusionSetTy *ExclusionSet = nullptr, 
-       SmallPtrSet<const Function *, 16> *Visited = nullptr) const = 0; 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAInterFnReachability &createForPosition(const IRPosition &IRP, 
-                                                   Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAInterFnReachability"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is AACallEdges. 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract interface for struct information. 
- struct AAPointerInfo : public AbstractAttribute { 
-   AAPointerInfo(const IRPosition &IRP) : AbstractAttribute(IRP) {} 
-   
-   enum AccessKind { 
-     // First two bits to distinguish may and must accesses. 
-     AK_MUST = 1 << 0, 
-     AK_MAY = 1 << 1, 
-   
-     // Then two bits for read and write. These are not exclusive. 
-     AK_R = 1 << 2, 
-     AK_W = 1 << 3, 
-     AK_RW = AK_R | AK_W, 
-   
-     // One special case for assumptions about memory content. These 
-     // are neither reads nor writes. They are however always modeled 
-     // as read to avoid using them for write removal. 
-     AK_ASSUMPTION = (1 << 4) | AK_MUST, 
-   
-     // Helper for easy access. 
-     AK_MAY_READ = AK_MAY | AK_R, 
-     AK_MAY_WRITE = AK_MAY | AK_W, 
-     AK_MAY_READ_WRITE = AK_MAY | AK_R | AK_W, 
-     AK_MUST_READ = AK_MUST | AK_R, 
-     AK_MUST_WRITE = AK_MUST | AK_W, 
-     AK_MUST_READ_WRITE = AK_MUST | AK_R | AK_W, 
-   }; 
-   
-   /// A container for a list of ranges. 
-   struct RangeList { 
-     // The set of ranges rarely contains more than one element, and is unlikely 
-     // to contain more than say four elements. So we find the middle-ground with 
-     // a sorted vector. This avoids hard-coding a rarely used number like "four" 
-     // into every instance of a SmallSet. 
-     using RangeTy = AA::RangeTy; 
-     using VecTy = SmallVector<RangeTy>; 
-     using iterator = VecTy::iterator; 
-     using const_iterator = VecTy::const_iterator; 
-     VecTy Ranges; 
-   
-     RangeList(const RangeTy &R) { Ranges.push_back(R); } 
-     RangeList(ArrayRef<int64_t> Offsets, int64_t Size) { 
-       Ranges.reserve(Offsets.size()); 
-       for (unsigned i = 0, e = Offsets.size(); i != e; ++i) { 
-         assert(((i + 1 == e) || Offsets[i] < Offsets[i + 1]) && 
-                "Expected strictly ascending offsets."); 
-         Ranges.emplace_back(Offsets[i], Size); 
-       } 
-     } 
-     RangeList() = default; 
-   
-     iterator begin() { return Ranges.begin(); } 
-     iterator end() { return Ranges.end(); } 
-     const_iterator begin() const { return Ranges.begin(); } 
-     const_iterator end() const { return Ranges.end(); } 
-   
-     // Helpers required for std::set_difference 
-     using value_type = RangeTy; 
-     void push_back(const RangeTy &R) { 
-       assert((Ranges.empty() || RangeTy::OffsetLessThan(Ranges.back(), R)) && 
-              "Ensure the last element is the greatest."); 
-       Ranges.push_back(R); 
-     } 
-   
-     /// Copy ranges from \p L that are not in \p R, into \p D. 
-     static void set_difference(const RangeList &L, const RangeList &R, 
-                                RangeList &D) { 
-       std::set_difference(L.begin(), L.end(), R.begin(), R.end(), 
-                           std::back_inserter(D), RangeTy::OffsetLessThan); 
-     } 
-   
-     unsigned size() const { return Ranges.size(); } 
-   
-     bool operator==(const RangeList &OI) const { return Ranges == OI.Ranges; } 
-   
-     /// Merge the ranges in \p RHS into the current ranges. 
-     /// - Merging a list of  unknown ranges makes the current list unknown. 
-     /// - Ranges with the same offset are merged according to RangeTy::operator& 
-     /// \return true if the current RangeList changed. 
-     bool merge(const RangeList &RHS) { 
-       if (isUnknown()) 
-         return false; 
-       if (RHS.isUnknown()) { 
-         setUnknown(); 
-         return true; 
-       } 
-   
-       if (Ranges.empty()) { 
-         Ranges = RHS.Ranges; 
-         return true; 
-       } 
-   
-       bool Changed = false; 
-       auto LPos = Ranges.begin(); 
-       for (auto &R : RHS.Ranges) { 
-         auto Result = insert(LPos, R); 
-         if (isUnknown()) 
-           return true; 
-         LPos = Result.first; 
-         Changed |= Result.second; 
-       } 
-       return Changed; 
-     } 
-   
-     /// Insert \p R at the given iterator \p Pos, and merge if necessary. 
-     /// 
-     /// This assumes that all ranges before \p Pos are OffsetLessThan \p R, and 
-     /// then maintains the sorted order for the suffix list. 
-     /// 
-     /// \return The place of insertion and true iff anything changed. 
-     std::pair<iterator, bool> insert(iterator Pos, const RangeTy &R) { 
-       if (isUnknown()) 
-         return std::make_pair(Ranges.begin(), false); 
-       if (R.offsetOrSizeAreUnknown()) { 
-         return std::make_pair(setUnknown(), true); 
-       } 
-   
-       // Maintain this as a sorted vector of unique entries. 
-       auto LB = std::lower_bound(Pos, Ranges.end(), R, RangeTy::OffsetLessThan); 
-       if (LB == Ranges.end() || LB->Offset != R.Offset) 
-         return std::make_pair(Ranges.insert(LB, R), true); 
-       bool Changed = *LB != R; 
-       *LB &= R; 
-       if (LB->offsetOrSizeAreUnknown()) 
-         return std::make_pair(setUnknown(), true); 
-       return std::make_pair(LB, Changed); 
-     } 
-   
-     /// Insert the given range \p R, maintaining sorted order. 
-     /// 
-     /// \return The place of insertion and true iff anything changed. 
-     std::pair<iterator, bool> insert(const RangeTy &R) { 
-       return insert(Ranges.begin(), R); 
-     } 
-   
-     /// Add the increment \p Inc to the offset of every range. 
-     void addToAllOffsets(int64_t Inc) { 
-       assert(!isUnassigned() && 
-              "Cannot increment if the offset is not yet computed!"); 
-       if (isUnknown()) 
-         return; 
-       for (auto &R : Ranges) { 
-         R.Offset += Inc; 
-       } 
-     } 
-   
-     /// Return true iff there is exactly one range and it is known. 
-     bool isUnique() const { 
-       return Ranges.size() == 1 && !Ranges.front().offsetOrSizeAreUnknown(); 
-     } 
-   
-     /// Return the unique range, assuming it exists. 
-     const RangeTy &getUnique() const { 
-       assert(isUnique() && "No unique range to return!"); 
-       return Ranges.front(); 
-     } 
-   
-     /// Return true iff the list contains an unknown range. 
-     bool isUnknown() const { 
-       if (isUnassigned()) 
-         return false; 
-       if (Ranges.front().offsetOrSizeAreUnknown()) { 
-         assert(Ranges.size() == 1 && "Unknown is a singleton range."); 
-         return true; 
-       } 
-       return false; 
-     } 
-   
-     /// Discard all ranges and insert a single unknown range. 
-     iterator setUnknown() { 
-       Ranges.clear(); 
-       Ranges.push_back(RangeTy::getUnknown()); 
-       return Ranges.begin(); 
-     } 
-   
-     /// Return true if no ranges have been inserted. 
-     bool isUnassigned() const { return Ranges.size() == 0; } 
-   }; 
-   
-   /// An access description. 
-   struct Access { 
-     Access(Instruction *I, int64_t Offset, int64_t Size, 
-            std::optional<Value *> Content, AccessKind Kind, Type *Ty) 
-         : LocalI(I), RemoteI(I), Content(Content), Ranges(Offset, Size), 
-           Kind(Kind), Ty(Ty) { 
-       verify(); 
-     } 
-     Access(Instruction *LocalI, Instruction *RemoteI, const RangeList &Ranges, 
-            std::optional<Value *> Content, AccessKind K, Type *Ty) 
-         : LocalI(LocalI), RemoteI(RemoteI), Content(Content), Ranges(Ranges), 
-           Kind(K), Ty(Ty) { 
-       if (Ranges.size() > 1) { 
-         Kind = AccessKind(Kind | AK_MAY); 
-         Kind = AccessKind(Kind & ~AK_MUST); 
-       } 
-       verify(); 
-     } 
-     Access(Instruction *LocalI, Instruction *RemoteI, int64_t Offset, 
-            int64_t Size, std::optional<Value *> Content, AccessKind Kind, 
-            Type *Ty) 
-         : LocalI(LocalI), RemoteI(RemoteI), Content(Content), 
-           Ranges(Offset, Size), Kind(Kind), Ty(Ty) { 
-       verify(); 
-     } 
-     Access(const Access &Other) = default; 
-   
-     Access &operator=(const Access &Other) = default; 
-     bool operator==(const Access &R) const { 
-       return LocalI == R.LocalI && RemoteI == R.RemoteI && Ranges == R.Ranges && 
-              Content == R.Content && Kind == R.Kind; 
-     } 
-     bool operator!=(const Access &R) const { return !(*this == R); } 
-   
-     Access &operator&=(const Access &R) { 
-       assert(RemoteI == R.RemoteI && "Expected same instruction!"); 
-       assert(LocalI == R.LocalI && "Expected same instruction!"); 
-   
-       // Note that every Access object corresponds to a unique Value, and only 
-       // accesses to the same Value are merged. Hence we assume that all ranges 
-       // are the same size. If ranges can be different size, then the contents 
-       // must be dropped. 
-       Ranges.merge(R.Ranges); 
-       Content = 
-           AA::combineOptionalValuesInAAValueLatice(Content, R.Content, Ty); 
-   
-       // Combine the access kind, which results in a bitwise union. 
-       // If there is more than one range, then this must be a MAY. 
-       // If we combine a may and a must access we clear the must bit. 
-       Kind = AccessKind(Kind | R.Kind); 
-       if ((Kind & AK_MAY) || Ranges.size() > 1) { 
-         Kind = AccessKind(Kind | AK_MAY); 
-         Kind = AccessKind(Kind & ~AK_MUST); 
-       } 
-       verify(); 
-       return *this; 
-     } 
-   
-     void verify() { 
-       assert(isMustAccess() + isMayAccess() == 1 && 
-              "Expect must or may access, not both."); 
-       assert(isAssumption() + isWrite() <= 1 && 
-              "Expect assumption access or write access, never both."); 
-       assert((isMayAccess() || Ranges.size() == 1) && 
-              "Cannot be a must access if there are multiple ranges."); 
-     } 
-   
-     /// Return the access kind. 
-     AccessKind getKind() const { return Kind; } 
-   
-     /// Return true if this is a read access. 
-     bool isRead() const { return Kind & AK_R; } 
-   
-     /// Return true if this is a write access. 
-     bool isWrite() const { return Kind & AK_W; } 
-   
-     /// Return true if this is a write access. 
-     bool isWriteOrAssumption() const { return isWrite() || isAssumption(); } 
-   
-     /// Return true if this is an assumption access. 
-     bool isAssumption() const { return Kind == AK_ASSUMPTION; } 
-   
-     bool isMustAccess() const { 
-       bool MustAccess = Kind & AK_MUST; 
-       assert((!MustAccess || Ranges.size() < 2) && 
-              "Cannot be a must access if there are multiple ranges."); 
-       return MustAccess; 
-     } 
-   
-     bool isMayAccess() const { 
-       bool MayAccess = Kind & AK_MAY; 
-       assert((MayAccess || Ranges.size() < 2) && 
-              "Cannot be a must access if there are multiple ranges."); 
-       return MayAccess; 
-     } 
-   
-     /// Return the instruction that causes the access with respect to the local 
-     /// scope of the associated attribute. 
-     Instruction *getLocalInst() const { return LocalI; } 
-   
-     /// Return the actual instruction that causes the access. 
-     Instruction *getRemoteInst() const { return RemoteI; } 
-   
-     /// Return true if the value written is not known yet. 
-     bool isWrittenValueYetUndetermined() const { return !Content; } 
-   
-     /// Return true if the value written cannot be determined at all. 
-     bool isWrittenValueUnknown() const { 
-       return Content.has_value() && !*Content; 
-     } 
-   
-     /// Set the value written to nullptr, i.e., unknown. 
-     void setWrittenValueUnknown() { Content = nullptr; } 
-   
-     /// Return the type associated with the access, if known. 
-     Type *getType() const { return Ty; } 
-   
-     /// Return the value writen, if any. 
-     Value *getWrittenValue() const { 
-       assert(!isWrittenValueYetUndetermined() && 
-              "Value needs to be determined before accessing it."); 
-       return *Content; 
-     } 
-   
-     /// Return the written value which can be `llvm::null` if it is not yet 
-     /// determined. 
-     std::optional<Value *> getContent() const { return Content; } 
-   
-     bool hasUniqueRange() const { return Ranges.isUnique(); } 
-     const AA::RangeTy &getUniqueRange() const { return Ranges.getUnique(); } 
-   
-     /// Add a range accessed by this Access. 
-     /// 
-     /// If there are multiple ranges, then this is a "may access". 
-     void addRange(int64_t Offset, int64_t Size) { 
-       Ranges.insert({Offset, Size}); 
-       if (!hasUniqueRange()) { 
-         Kind = AccessKind(Kind | AK_MAY); 
-         Kind = AccessKind(Kind & ~AK_MUST); 
-       } 
-     } 
-   
-     const RangeList &getRanges() const { return Ranges; } 
-   
-     using const_iterator = RangeList::const_iterator; 
-     const_iterator begin() const { return Ranges.begin(); } 
-     const_iterator end() const { return Ranges.end(); } 
-   
-   private: 
-     /// The instruction responsible for the access with respect to the local 
-     /// scope of the associated attribute. 
-     Instruction *LocalI; 
-   
-     /// The instruction responsible for the access. 
-     Instruction *RemoteI; 
-   
-     /// The value written, if any. `llvm::none` means "not known yet", `nullptr` 
-     /// cannot be determined. 
-     std::optional<Value *> Content; 
-   
-     /// Set of potential ranges accessed from the base pointer. 
-     RangeList Ranges; 
-   
-     /// The access kind, e.g., READ, as bitset (could be more than one). 
-     AccessKind Kind; 
-   
-     /// The type of the content, thus the type read/written, can be null if not 
-     /// available. 
-     Type *Ty; 
-   }; 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAPointerInfo &createForPosition(const IRPosition &IRP, Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAPointerInfo"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// Call \p CB on all accesses that might interfere with \p Range and return 
-   /// true if all such accesses were known and the callback returned true for 
-   /// all of them, false otherwise. An access interferes with an offset-size 
-   /// pair if it might read or write that memory region. 
-   virtual bool forallInterferingAccesses( 
-       AA::RangeTy Range, function_ref<bool(const Access &, bool)> CB) const = 0; 
-   
-   /// Call \p CB on all accesses that might interfere with \p I and 
-   /// return true if all such accesses were known and the callback returned true 
-   /// for all of them, false otherwise. In contrast to forallInterferingAccesses 
-   /// this function will perform reasoning to exclude write accesses that cannot 
-   /// affect the load even if they on the surface look as if they would. The 
-   /// flag \p HasBeenWrittenTo will be set to true if we know that \p I does not 
-   /// read the intial value of the underlying memory. 
-   virtual bool forallInterferingAccesses( 
-       Attributor &A, const AbstractAttribute &QueryingAA, Instruction &I, 
-       function_ref<bool(const Access &, bool)> CB, bool &HasBeenWrittenTo, 
-       AA::RangeTy &Range) const = 0; 
-   
-   /// This function should return true if the type of the \p AA is AAPointerInfo 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract attribute for getting assumption information. 
- struct AAAssumptionInfo 
-     : public StateWrapper<SetState<StringRef>, AbstractAttribute, 
-                           DenseSet<StringRef>> { 
-   using Base = 
-       StateWrapper<SetState<StringRef>, AbstractAttribute, DenseSet<StringRef>>; 
-   
-   AAAssumptionInfo(const IRPosition &IRP, Attributor &A, 
-                    const DenseSet<StringRef> &Known) 
-       : Base(IRP, Known) {} 
-   
-   /// Returns true if the assumption set contains the assumption \p Assumption. 
-   virtual bool hasAssumption(const StringRef Assumption) const = 0; 
-   
-   /// Create an abstract attribute view for the position \p IRP. 
-   static AAAssumptionInfo &createForPosition(const IRPosition &IRP, 
-                                              Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAAssumptionInfo"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAAssumptionInfo 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
- }; 
-   
- /// An abstract attribute for getting all assumption underlying objects. 
- struct AAUnderlyingObjects : AbstractAttribute { 
-   AAUnderlyingObjects(const IRPosition &IRP) : AbstractAttribute(IRP) {} 
-   
-   /// Create an abstract attribute biew for the position \p IRP. 
-   static AAUnderlyingObjects &createForPosition(const IRPosition &IRP, 
-                                                 Attributor &A); 
-   
-   /// See AbstractAttribute::getName() 
-   const std::string getName() const override { return "AAUnderlyingObjects"; } 
-   
-   /// See AbstractAttribute::getIdAddr() 
-   const char *getIdAddr() const override { return &ID; } 
-   
-   /// This function should return true if the type of the \p AA is 
-   /// AAUnderlyingObjects. 
-   static bool classof(const AbstractAttribute *AA) { 
-     return (AA->getIdAddr() == &ID); 
-   } 
-   
-   /// Unique ID (due to the unique address) 
-   static const char ID; 
-   
-   /// Check \p Pred on all underlying objects in \p Scope collected so far. 
-   /// 
-   /// This method will evaluate \p Pred on all underlying objects in \p Scope 
-   /// collected so far and return true if \p Pred holds on all of them. 
-   virtual bool 
-   forallUnderlyingObjects(function_ref<bool(Value &)> Pred, 
-                           AA::ValueScope Scope = AA::Interprocedural) const = 0; 
- }; 
-   
- raw_ostream &operator<<(raw_ostream &, const AAPointerInfo::Access &); 
-   
- /// Run options, used by the pass manager. 
- enum AttributorRunOption { 
-   NONE = 0, 
-   MODULE = 1 << 0, 
-   CGSCC = 1 << 1, 
-   ALL = MODULE | CGSCC 
- }; 
-   
- } // end namespace llvm 
-   
- #endif // LLVM_TRANSFORMS_IPO_ATTRIBUTOR_H 
-