InlineFunction.cpp

//===- InlineFunction.cpp - Code to perform function inlining -------------===//
//
//                     The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file implements inlining of a function into a call site, resolving
// parameters and the return value as appropriate.
//
//===----------------------------------------------------------------------===//

#include "llvm/Transforms/Utils/Cloning.h"
#include "llvm/ADT/SmallSet.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/SetVector.h"
#include "llvm/ADT/StringExtras.h"
#include "llvm/Analysis/AliasAnalysis.h"
#include "llvm/Analysis/AssumptionTracker.h"
#include "llvm/Analysis/CallGraph.h"
#include "llvm/Analysis/CaptureTracking.h"
#include "llvm/Analysis/InstructionSimplify.h"
#include "llvm/Analysis/ValueTracking.h"
#include "llvm/IR/Attributes.h"
#include "llvm/IR/CallSite.h"
#include "llvm/IR/CFG.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/DebugInfo.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/IRBuilder.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/Intrinsics.h"
#include "llvm/IR/MDBuilder.h"
#include "llvm/IR/Module.h"
#include "llvm/Transforms/Utils/Local.h"
#include "llvm/Support/CommandLine.h"
#include <algorithm>
using namespace llvm;

static cl::opt<bool>
EnableNoAliasConversion("enable-noalias-to-md-conversion", cl::init(true),
  cl::Hidden,
  cl::desc("Convert noalias attributes to metadata during inlining."));

static cl::opt<bool>
PreserveAlignmentAssumptions("preserve-alignment-assumptions-during-inlining",
  cl::init(true), cl::Hidden,
  cl::desc("Convert align attributes to assumptions during inlining."));

bool llvm::InlineFunction(CallInst *CI, InlineFunctionInfo &IFI,
                          bool InsertLifetime) {
  return InlineFunction(CallSite(CI), IFI, InsertLifetime);
}
bool llvm::InlineFunction(InvokeInst *II, InlineFunctionInfo &IFI,
                          bool InsertLifetime) {
  return InlineFunction(CallSite(II), IFI, InsertLifetime);
}

namespace {
  /// A class for recording information about inlining through an invoke.
  class InvokeInliningInfo {
    BasicBlock *OuterResumeDest; ///< Destination of the invoke's unwind.
    BasicBlock *InnerResumeDest; ///< Destination for the callee's resume.
    LandingPadInst *CallerLPad;  ///< LandingPadInst associated with the invoke.
    PHINode *InnerEHValuesPHI;   ///< PHI for EH values from landingpad insts.
    SmallVector<Value*, 8> UnwindDestPHIValues;

  public:
    InvokeInliningInfo(InvokeInst *II)
      : OuterResumeDest(II->getUnwindDest()), InnerResumeDest(nullptr),
        CallerLPad(nullptr), InnerEHValuesPHI(nullptr) {
      // If there are PHI nodes in the unwind destination block, we need to keep
      // track of which values came into them from the invoke before removing
      // the edge from this block.
      llvm::BasicBlock *InvokeBB = II->getParent();
      BasicBlock::iterator I = OuterResumeDest->begin();
      for (; isa<PHINode>(I); ++I) {
        // Save the value to use for this edge.
        PHINode *PHI = cast<PHINode>(I);
        UnwindDestPHIValues.push_back(PHI->getIncomingValueForBlock(InvokeBB));
      }

      CallerLPad = cast<LandingPadInst>(I);
    }

    /// getOuterResumeDest - The outer unwind destination is the target of
    /// unwind edges introduced for calls within the inlined function.
    BasicBlock *getOuterResumeDest() const {
      return OuterResumeDest;
    }

    BasicBlock *getInnerResumeDest();

    LandingPadInst *getLandingPadInst() const { return CallerLPad; }

    /// forwardResume - Forward the 'resume' instruction to the caller's landing
    /// pad block. When the landing pad block has only one predecessor, this is
    /// a simple branch. When there is more than one predecessor, we need to
    /// split the landing pad block after the landingpad instruction and jump
    /// to there.
    void forwardResume(ResumeInst *RI,
                       SmallPtrSetImpl<LandingPadInst*> &InlinedLPads);

    /// addIncomingPHIValuesFor - Add incoming-PHI values to the unwind
    /// destination block for the given basic block, using the values for the
    /// original invoke's source block.
    void addIncomingPHIValuesFor(BasicBlock *BB) const {
      addIncomingPHIValuesForInto(BB, OuterResumeDest);
    }

    void addIncomingPHIValuesForInto(BasicBlock *src, BasicBlock *dest) const {
      BasicBlock::iterator I = dest->begin();
      for (unsigned i = 0, e = UnwindDestPHIValues.size(); i != e; ++i, ++I) {
        PHINode *phi = cast<PHINode>(I);
        phi->addIncoming(UnwindDestPHIValues[i], src);
      }
    }
  };
}

/// getInnerResumeDest - Get or create a target for the branch from ResumeInsts.
BasicBlock *InvokeInliningInfo::getInnerResumeDest() {
  if (InnerResumeDest) return InnerResumeDest;

  // Split the landing pad.
  BasicBlock::iterator SplitPoint = CallerLPad; ++SplitPoint;
  InnerResumeDest =
    OuterResumeDest->splitBasicBlock(SplitPoint,
                                     OuterResumeDest->getName() + ".body");

  // The number of incoming edges we expect to the inner landing pad.
  const unsigned PHICapacity = 2;

  // Create corresponding new PHIs for all the PHIs in the outer landing pad.
  BasicBlock::iterator InsertPoint = InnerResumeDest->begin();
  BasicBlock::iterator I = OuterResumeDest->begin();
  for (unsigned i = 0, e = UnwindDestPHIValues.size(); i != e; ++i, ++I) {
    PHINode *OuterPHI = cast<PHINode>(I);
    PHINode *InnerPHI = PHINode::Create(OuterPHI->getType(), PHICapacity,
                                        OuterPHI->getName() + ".lpad-body",
                                        InsertPoint);
    OuterPHI->replaceAllUsesWith(InnerPHI);
    InnerPHI->addIncoming(OuterPHI, OuterResumeDest);
  }

  // Create a PHI for the exception values.
  InnerEHValuesPHI = PHINode::Create(CallerLPad->getType(), PHICapacity,
                                     "eh.lpad-body", InsertPoint);
  CallerLPad->replaceAllUsesWith(InnerEHValuesPHI);
  InnerEHValuesPHI->addIncoming(CallerLPad, OuterResumeDest);

  // All done.
  return InnerResumeDest;
}

/// forwardResume - Forward the 'resume' instruction to the caller's landing pad
/// block. When the landing pad block has only one predecessor, this is a simple
/// branch. When there is more than one predecessor, we need to split the
/// landing pad block after the landingpad instruction and jump to there.
void InvokeInliningInfo::forwardResume(ResumeInst *RI,
                               SmallPtrSetImpl<LandingPadInst*> &InlinedLPads) {
  BasicBlock *Dest = getInnerResumeDest();
  BasicBlock *Src = RI->getParent();

  BranchInst::Create(Dest, Src);

  // Update the PHIs in the destination. They were inserted in an order which
  // makes this work.
  addIncomingPHIValuesForInto(Src, Dest);

  InnerEHValuesPHI->addIncoming(RI->getOperand(0), Src);
  RI->eraseFromParent();
}

/// HandleCallsInBlockInlinedThroughInvoke - When we inline a basic block into
/// an invoke, we have to turn all of the calls that can throw into
/// invokes.  This function analyze BB to see if there are any calls, and if so,
/// it rewrites them to be invokes that jump to InvokeDest and fills in the PHI
/// nodes in that block with the values specified in InvokeDestPHIValues.
static void HandleCallsInBlockInlinedThroughInvoke(BasicBlock *BB,
                                                   InvokeInliningInfo &Invoke) {
  for (BasicBlock::iterator BBI = BB->begin(), E = BB->end(); BBI != E; ) {
    Instruction *I = BBI++;

    // We only need to check for function calls: inlined invoke
    // instructions require no special handling.
    CallInst *CI = dyn_cast<CallInst>(I);

    // If this call cannot unwind, don't convert it to an invoke.
    // Inline asm calls cannot throw.
    if (!CI || CI->doesNotThrow() || isa<InlineAsm>(CI->getCalledValue()))
      continue;

    // Convert this function call into an invoke instruction.  First, split the
    // basic block.
    BasicBlock *Split = BB->splitBasicBlock(CI, CI->getName()+".noexc");

    // Delete the unconditional branch inserted by splitBasicBlock
    BB->getInstList().pop_back();

    // Create the new invoke instruction.
    ImmutableCallSite CS(CI);
    SmallVector<Value*, 8> InvokeArgs(CS.arg_begin(), CS.arg_end());
    InvokeInst *II = InvokeInst::Create(CI->getCalledValue(), Split,
                                        Invoke.getOuterResumeDest(),
                                        InvokeArgs, CI->getName(), BB);
    II->setDebugLoc(CI->getDebugLoc());
    II->setCallingConv(CI->getCallingConv());
    II->setAttributes(CI->getAttributes());
    
    // Make sure that anything using the call now uses the invoke!  This also
    // updates the CallGraph if present, because it uses a WeakVH.
    CI->replaceAllUsesWith(II);

    // Delete the original call
    Split->getInstList().pop_front();

    // Update any PHI nodes in the exceptional block to indicate that there is
    // now a new entry in them.
    Invoke.addIncomingPHIValuesFor(BB);
    return;
  }
}

/// HandleInlinedInvoke - If we inlined an invoke site, we need to convert calls
/// in the body of the inlined function into invokes.
///
/// II is the invoke instruction being inlined.  FirstNewBlock is the first
/// block of the inlined code (the last block is the end of the function),
/// and InlineCodeInfo is information about the code that got inlined.
static void HandleInlinedInvoke(InvokeInst *II, BasicBlock *FirstNewBlock,
                                ClonedCodeInfo &InlinedCodeInfo) {
  BasicBlock *InvokeDest = II->getUnwindDest();

  Function *Caller = FirstNewBlock->getParent();

  // The inlined code is currently at the end of the function, scan from the
  // start of the inlined code to its end, checking for stuff we need to
  // rewrite.
  InvokeInliningInfo Invoke(II);

  // Get all of the inlined landing pad instructions.
  SmallPtrSet<LandingPadInst*, 16> InlinedLPads;
  for (Function::iterator I = FirstNewBlock, E = Caller->end(); I != E; ++I)
    if (InvokeInst *II = dyn_cast<InvokeInst>(I->getTerminator()))
      InlinedLPads.insert(II->getLandingPadInst());

  // Append the clauses from the outer landing pad instruction into the inlined
  // landing pad instructions.
  LandingPadInst *OuterLPad = Invoke.getLandingPadInst();
  for (LandingPadInst *InlinedLPad : InlinedLPads) {
    unsigned OuterNum = OuterLPad->getNumClauses();
    InlinedLPad->reserveClauses(OuterNum);
    for (unsigned OuterIdx = 0; OuterIdx != OuterNum; ++OuterIdx)
      InlinedLPad->addClause(OuterLPad->getClause(OuterIdx));
    if (OuterLPad->isCleanup())
      InlinedLPad->setCleanup(true);
  }

  for (Function::iterator BB = FirstNewBlock, E = Caller->end(); BB != E; ++BB){
    if (InlinedCodeInfo.ContainsCalls)
      HandleCallsInBlockInlinedThroughInvoke(BB, Invoke);

    // Forward any resumes that are remaining here.
    if (ResumeInst *RI = dyn_cast<ResumeInst>(BB->getTerminator()))
      Invoke.forwardResume(RI, InlinedLPads);
  }

  // Now that everything is happy, we have one final detail.  The PHI nodes in
  // the exception destination block still have entries due to the original
  // invoke instruction. Eliminate these entries (which might even delete the
  // PHI node) now.
  InvokeDest->removePredecessor(II->getParent());
}

/// CloneAliasScopeMetadata - When inlining a function that contains noalias
/// scope metadata, this metadata needs to be cloned so that the inlined blocks
/// have different "unqiue scopes" at every call site. Were this not done, then
/// aliasing scopes from a function inlined into a caller multiple times could
/// not be differentiated (and this would lead to miscompiles because the
/// non-aliasing property communicated by the metadata could have
/// call-site-specific control dependencies).
static void CloneAliasScopeMetadata(CallSite CS, ValueToValueMapTy &VMap) {
  const Function *CalledFunc = CS.getCalledFunction();
  SetVector<const MDNode *> MD;

  // Note: We could only clone the metadata if it is already used in the
  // caller. I'm omitting that check here because it might confuse
  // inter-procedural alias analysis passes. We can revisit this if it becomes
  // an efficiency or overhead problem.

  for (Function::const_iterator I = CalledFunc->begin(), IE = CalledFunc->end();
       I != IE; ++I)
    for (BasicBlock::const_iterator J = I->begin(), JE = I->end(); J != JE; ++J) {
      if (const MDNode *M = J->getMetadata(LLVMContext::MD_alias_scope))
        MD.insert(M);
      if (const MDNode *M = J->getMetadata(LLVMContext::MD_noalias))
        MD.insert(M);
    }

  if (MD.empty())
    return;

  // Walk the existing metadata, adding the complete (perhaps cyclic) chain to
  // the set.
  SmallVector<const Value *, 16> Queue(MD.begin(), MD.end());
  while (!Queue.empty()) {
    const MDNode *M = cast<MDNode>(Queue.pop_back_val());
    for (unsigned i = 0, ie = M->getNumOperands(); i != ie; ++i)
      if (const MDNode *M1 = dyn_cast<MDNode>(M->getOperand(i)))
        if (MD.insert(M1))
          Queue.push_back(M1);
  }

  // Now we have a complete set of all metadata in the chains used to specify
  // the noalias scopes and the lists of those scopes.
  SmallVector<MDNode *, 16> DummyNodes;
  DenseMap<const MDNode *, TrackingVH<MDNode> > MDMap;
  for (SetVector<const MDNode *>::iterator I = MD.begin(), IE = MD.end();
       I != IE; ++I) {
    MDNode *Dummy = MDNode::getTemporary(CalledFunc->getContext(), None);
    DummyNodes.push_back(Dummy);
    MDMap[*I] = Dummy;
  }

  // Create new metadata nodes to replace the dummy nodes, replacing old
  // metadata references with either a dummy node or an already-created new
  // node.
  for (SetVector<const MDNode *>::iterator I = MD.begin(), IE = MD.end();
       I != IE; ++I) {
    SmallVector<Value *, 4> NewOps;
    for (unsigned i = 0, ie = (*I)->getNumOperands(); i != ie; ++i) {
      const Value *V = (*I)->getOperand(i);
      if (const MDNode *M = dyn_cast<MDNode>(V))
        NewOps.push_back(MDMap[M]);
      else
        NewOps.push_back(const_cast<Value *>(V));
    }

    MDNode *NewM = MDNode::get(CalledFunc->getContext(), NewOps),
           *TempM = MDMap[*I];

    TempM->replaceAllUsesWith(NewM);
  }

  // Now replace the metadata in the new inlined instructions with the
  // repacements from the map.
  for (ValueToValueMapTy::iterator VMI = VMap.begin(), VMIE = VMap.end();
       VMI != VMIE; ++VMI) {
    if (!VMI->second)
      continue;

    Instruction *NI = dyn_cast<Instruction>(VMI->second);
    if (!NI)
      continue;

    if (MDNode *M = NI->getMetadata(LLVMContext::MD_alias_scope)) {
      MDNode *NewMD = MDMap[M];
      // If the call site also had alias scope metadata (a list of scopes to
      // which instructions inside it might belong), propagate those scopes to
      // the inlined instructions.
      if (MDNode *CSM =
          CS.getInstruction()->getMetadata(LLVMContext::MD_alias_scope))
        NewMD = MDNode::concatenate(NewMD, CSM);
      NI->setMetadata(LLVMContext::MD_alias_scope, NewMD);
    } else if (NI->mayReadOrWriteMemory()) {
      if (MDNode *M =
          CS.getInstruction()->getMetadata(LLVMContext::MD_alias_scope))
        NI->setMetadata(LLVMContext::MD_alias_scope, M);
    }

    if (MDNode *M = NI->getMetadata(LLVMContext::MD_noalias)) {
      MDNode *NewMD = MDMap[M];
      // If the call site also had noalias metadata (a list of scopes with
      // which instructions inside it don't alias), propagate those scopes to
      // the inlined instructions.
      if (MDNode *CSM =
          CS.getInstruction()->getMetadata(LLVMContext::MD_noalias))
        NewMD = MDNode::concatenate(NewMD, CSM);
      NI->setMetadata(LLVMContext::MD_noalias, NewMD);
    } else if (NI->mayReadOrWriteMemory()) {
      if (MDNode *M =
          CS.getInstruction()->getMetadata(LLVMContext::MD_noalias))
        NI->setMetadata(LLVMContext::MD_noalias, M);
    }
  }

  // Now that everything has been replaced, delete the dummy nodes.
  for (unsigned i = 0, ie = DummyNodes.size(); i != ie; ++i)
    MDNode::deleteTemporary(DummyNodes[i]);
}

/// AddAliasScopeMetadata - If the inlined function has noalias arguments, then
/// add new alias scopes for each noalias argument, tag the mapped noalias
/// parameters with noalias metadata specifying the new scope, and tag all
/// non-derived loads, stores and memory intrinsics with the new alias scopes.
static void AddAliasScopeMetadata(CallSite CS, ValueToValueMapTy &VMap,
                                  const DataLayout *DL, AliasAnalysis *AA) {
  if (!EnableNoAliasConversion)
    return;

  const Function *CalledFunc = CS.getCalledFunction();
  SmallVector<const Argument *, 4> NoAliasArgs;

  for (Function::const_arg_iterator I = CalledFunc->arg_begin(),
       E = CalledFunc->arg_end(); I != E; ++I) {
    if (I->hasNoAliasAttr() && !I->hasNUses(0))
      NoAliasArgs.push_back(I);
  }

  if (NoAliasArgs.empty())
    return;

  // To do a good job, if a noalias variable is captured, we need to know if
  // the capture point dominates the particular use we're considering.
  DominatorTree DT;
  DT.recalculate(const_cast<Function&>(*CalledFunc));

  // noalias indicates that pointer values based on the argument do not alias
  // pointer values which are not based on it. So we add a new "scope" for each
  // noalias function argument. Accesses using pointers based on that argument
  // become part of that alias scope, accesses using pointers not based on that
  // argument are tagged as noalias with that scope.

  DenseMap<const Argument *, MDNode *> NewScopes;
  MDBuilder MDB(CalledFunc->getContext());

  // Create a new scope domain for this function.
  MDNode *NewDomain =
    MDB.createAnonymousAliasScopeDomain(CalledFunc->getName());
  for (unsigned i = 0, e = NoAliasArgs.size(); i != e; ++i) {
    const Argument *A = NoAliasArgs[i];

    std::string Name = CalledFunc->getName();
    if (A->hasName()) {
      Name += ": %";
      Name += A->getName();
    } else {
      Name += ": argument ";
      Name += utostr(i);
    }

    // Note: We always create a new anonymous root here. This is true regardless
    // of the linkage of the callee because the aliasing "scope" is not just a
    // property of the callee, but also all control dependencies in the caller.
    MDNode *NewScope = MDB.createAnonymousAliasScope(NewDomain, Name);
    NewScopes.insert(std::make_pair(A, NewScope));
  }

  // Iterate over all new instructions in the map; for all memory-access
  // instructions, add the alias scope metadata.
  for (ValueToValueMapTy::iterator VMI = VMap.begin(), VMIE = VMap.end();
       VMI != VMIE; ++VMI) {
    if (const Instruction *I = dyn_cast<Instruction>(VMI->first)) {
      if (!VMI->second)
        continue;

      Instruction *NI = dyn_cast<Instruction>(VMI->second);
      if (!NI)
        continue;

      bool IsArgMemOnlyCall = false, IsFuncCall = false;
      SmallVector<const Value *, 2> PtrArgs;

      if (const LoadInst *LI = dyn_cast<LoadInst>(I))
        PtrArgs.push_back(LI->getPointerOperand());
      else if (const StoreInst *SI = dyn_cast<StoreInst>(I))
        PtrArgs.push_back(SI->getPointerOperand());
      else if (const VAArgInst *VAAI = dyn_cast<VAArgInst>(I))
        PtrArgs.push_back(VAAI->getPointerOperand());
      else if (const AtomicCmpXchgInst *CXI = dyn_cast<AtomicCmpXchgInst>(I))
        PtrArgs.push_back(CXI->getPointerOperand());
      else if (const AtomicRMWInst *RMWI = dyn_cast<AtomicRMWInst>(I))
        PtrArgs.push_back(RMWI->getPointerOperand());
      else if (ImmutableCallSite ICS = ImmutableCallSite(I)) {
        // If we know that the call does not access memory, then we'll still
        // know that about the inlined clone of this call site, and we don't
        // need to add metadata.
        if (ICS.doesNotAccessMemory())
          continue;

        IsFuncCall = true;
        if (AA) {
          AliasAnalysis::ModRefBehavior MRB = AA->getModRefBehavior(ICS);
          if (MRB == AliasAnalysis::OnlyAccessesArgumentPointees ||
              MRB == AliasAnalysis::OnlyReadsArgumentPointees)
            IsArgMemOnlyCall = true;
        }

        for (ImmutableCallSite::arg_iterator AI = ICS.arg_begin(),
             AE = ICS.arg_end(); AI != AE; ++AI) {
          // We need to check the underlying objects of all arguments, not just
          // the pointer arguments, because we might be passing pointers as
          // integers, etc.
          // However, if we know that the call only accesses pointer arguments,
          // then we only need to check the pointer arguments.
          if (IsArgMemOnlyCall && !(*AI)->getType()->isPointerTy())
            continue;

          PtrArgs.push_back(*AI);
        }
      }

      // If we found no pointers, then this instruction is not suitable for
      // pairing with an instruction to receive aliasing metadata.
      // However, if this is a call, this we might just alias with none of the
      // noalias arguments.
      if (PtrArgs.empty() && !IsFuncCall)
        continue;

      // It is possible that there is only one underlying object, but you
      // need to go through several PHIs to see it, and thus could be
      // repeated in the Objects list.
      SmallPtrSet<const Value *, 4> ObjSet;
      SmallVector<Value *, 4> Scopes, NoAliases;

      SmallSetVector<const Argument *, 4> NAPtrArgs;
      for (unsigned i = 0, ie = PtrArgs.size(); i != ie; ++i) {
        SmallVector<Value *, 4> Objects;
        GetUnderlyingObjects(const_cast<Value*>(PtrArgs[i]),
                             Objects, DL, /* MaxLookup = */ 0);

        for (Value *O : Objects)
          ObjSet.insert(O);
      }

      // Figure out if we're derived from anything that is not a noalias
      // argument.
      bool CanDeriveViaCapture = false, UsesAliasingPtr = false;
      for (const Value *V : ObjSet) {
        // Is this value a constant that cannot be derived from any pointer
        // value (we need to exclude constant expressions, for example, that
        // are formed from arithmetic on global symbols).
        bool IsNonPtrConst = isa<ConstantInt>(V) || isa<ConstantFP>(V) ||
                             isa<ConstantPointerNull>(V) ||
                             isa<ConstantDataVector>(V) || isa<UndefValue>(V);
        if (IsNonPtrConst)
          continue;

        // If this is anything other than a noalias argument, then we cannot
        // completely describe the aliasing properties using alias.scope
        // metadata (and, thus, won't add any).
        if (const Argument *A = dyn_cast<Argument>(V)) {
          if (!A->hasNoAliasAttr())
            UsesAliasingPtr = true;
        } else {
          UsesAliasingPtr = true;
        }

        // If this is not some identified function-local object (which cannot
        // directly alias a noalias argument), or some other argument (which,
        // by definition, also cannot alias a noalias argument), then we could
        // alias a noalias argument that has been captured).
        if (!isa<Argument>(V) &&
            !isIdentifiedFunctionLocal(const_cast<Value*>(V)))
          CanDeriveViaCapture = true;
      }

      // A function call can always get captured noalias pointers (via other
      // parameters, globals, etc.).
      if (IsFuncCall && !IsArgMemOnlyCall)
        CanDeriveViaCapture = true;

      // First, we want to figure out all of the sets with which we definitely
      // don't alias. Iterate over all noalias set, and add those for which:
      //   1. The noalias argument is not in the set of objects from which we
      //      definitely derive.
      //   2. The noalias argument has not yet been captured.
      // An arbitrary function that might load pointers could see captured
      // noalias arguments via other noalias arguments or globals, and so we
      // must always check for prior capture.
      for (const Argument *A : NoAliasArgs) {
        if (!ObjSet.count(A) && (!CanDeriveViaCapture ||
                                 // It might be tempting to skip the
                                 // PointerMayBeCapturedBefore check if
                                 // A->hasNoCaptureAttr() is true, but this is
                                 // incorrect because nocapture only guarantees
                                 // that no copies outlive the function, not
                                 // that the value cannot be locally captured.
                                 !PointerMayBeCapturedBefore(A,
                                   /* ReturnCaptures */ false,
                                   /* StoreCaptures */ false, I, &DT)))
          NoAliases.push_back(NewScopes[A]);
      }

      if (!NoAliases.empty())
        NI->setMetadata(LLVMContext::MD_noalias, MDNode::concatenate(
          NI->getMetadata(LLVMContext::MD_noalias),
            MDNode::get(CalledFunc->getContext(), NoAliases)));

      // Next, we want to figure out all of the sets to which we might belong.
      // We might belong to a set if the noalias argument is in the set of
      // underlying objects. If there is some non-noalias argument in our list
      // of underlying objects, then we cannot add a scope because the fact
      // that some access does not alias with any set of our noalias arguments
      // cannot itself guarantee that it does not alias with this access
      // (because there is some pointer of unknown origin involved and the
      // other access might also depend on this pointer). We also cannot add
      // scopes to arbitrary functions unless we know they don't access any
      // non-parameter pointer-values.
      bool CanAddScopes = !UsesAliasingPtr;
      if (CanAddScopes && IsFuncCall)
        CanAddScopes = IsArgMemOnlyCall;

      if (CanAddScopes)
        for (const Argument *A : NoAliasArgs) {
          if (ObjSet.count(A))
            Scopes.push_back(NewScopes[A]);
        }

      if (!Scopes.empty())
        NI->setMetadata(LLVMContext::MD_alias_scope, MDNode::concatenate(
          NI->getMetadata(LLVMContext::MD_alias_scope),
            MDNode::get(CalledFunc->getContext(), Scopes)));
    }
  }
}

/// If the inlined function has non-byval align arguments, then
/// add @llvm.assume-based alignment assumptions to preserve this information.
static void AddAlignmentAssumptions(CallSite CS, InlineFunctionInfo &IFI) {
  if (!PreserveAlignmentAssumptions || !IFI.DL)
    return;

  // To avoid inserting redundant assumptions, we should check for assumptions
  // already in the caller. To do this, we might need a DT of the caller.
  DominatorTree DT;
  bool DTCalculated = false;

  const Function *CalledFunc = CS.getCalledFunction();
  for (Function::const_arg_iterator I = CalledFunc->arg_begin(),
       E = CalledFunc->arg_end(); I != E; ++I) {
    unsigned Align = I->getType()->isPointerTy() ? I->getParamAlignment() : 0;
    if (Align && !I->hasByValOrInAllocaAttr() && !I->hasNUses(0)) {
      if (!DTCalculated) {
        DT.recalculate(const_cast<Function&>(*CS.getInstruction()->getParent()
                                               ->getParent()));
        DTCalculated = true;
      }

      // If we can already prove the asserted alignment in the context of the
      // caller, then don't bother inserting the assumption.
      Value *Arg = CS.getArgument(I->getArgNo());
      if (getKnownAlignment(Arg, IFI.DL, IFI.AT, CS.getInstruction(),
                            &DT) >= Align)
        continue;

      IRBuilder<>(CS.getInstruction()).CreateAlignmentAssumption(*IFI.DL, Arg,
                                                                 Align);
    }
  }
}

/// UpdateCallGraphAfterInlining - Once we have cloned code over from a callee
/// into the caller, update the specified callgraph to reflect the changes we
/// made.  Note that it's possible that not all code was copied over, so only
/// some edges of the callgraph may remain.
static void UpdateCallGraphAfterInlining(CallSite CS,
                                         Function::iterator FirstNewBlock,
                                         ValueToValueMapTy &VMap,
                                         InlineFunctionInfo &IFI) {
  CallGraph &CG = *IFI.CG;
  const Function *Caller = CS.getInstruction()->getParent()->getParent();
  const Function *Callee = CS.getCalledFunction();
  CallGraphNode *CalleeNode = CG[Callee];
  CallGraphNode *CallerNode = CG[Caller];

  // Since we inlined some uninlined call sites in the callee into the caller,
  // add edges from the caller to all of the callees of the callee.
  CallGraphNode::iterator I = CalleeNode->begin(), E = CalleeNode->end();

  // Consider the case where CalleeNode == CallerNode.
  CallGraphNode::CalledFunctionsVector CallCache;
  if (CalleeNode == CallerNode) {
    CallCache.assign(I, E);
    I = CallCache.begin();
    E = CallCache.end();
  }

  for (; I != E; ++I) {
    const Value *OrigCall = I->first;

    ValueToValueMapTy::iterator VMI = VMap.find(OrigCall);
    // Only copy the edge if the call was inlined!
    if (VMI == VMap.end() || VMI->second == nullptr)
      continue;
    
    // If the call was inlined, but then constant folded, there is no edge to
    // add.  Check for this case.
    Instruction *NewCall = dyn_cast<Instruction>(VMI->second);
    if (!NewCall) continue;

    // Remember that this call site got inlined for the client of
    // InlineFunction.
    IFI.InlinedCalls.push_back(NewCall);

    // It's possible that inlining the callsite will cause it to go from an
    // indirect to a direct call by resolving a function pointer.  If this
    // happens, set the callee of the new call site to a more precise
    // destination.  This can also happen if the call graph node of the caller
    // was just unnecessarily imprecise.
    if (!I->second->getFunction())
      if (Function *F = CallSite(NewCall).getCalledFunction()) {
        // Indirect call site resolved to direct call.
        CallerNode->addCalledFunction(CallSite(NewCall), CG[F]);

        continue;
      }

    CallerNode->addCalledFunction(CallSite(NewCall), I->second);
  }
  
  // Update the call graph by deleting the edge from Callee to Caller.  We must
  // do this after the loop above in case Caller and Callee are the same.
  CallerNode->removeCallEdgeFor(CS);
}

static void HandleByValArgumentInit(Value *Dst, Value *Src, Module *M,
                                    BasicBlock *InsertBlock,
                                    InlineFunctionInfo &IFI) {
  Type *AggTy = cast<PointerType>(Src->getType())->getElementType();
  IRBuilder<> Builder(InsertBlock->begin());

  Value *Size;
  if (IFI.DL == nullptr)
    Size = ConstantExpr::getSizeOf(AggTy);
  else
    Size = Builder.getInt64(IFI.DL->getTypeStoreSize(AggTy));

  // Always generate a memcpy of alignment 1 here because we don't know
  // the alignment of the src pointer.  Other optimizations can infer
  // better alignment.
  Builder.CreateMemCpy(Dst, Src, Size, /*Align=*/1);
}

/// HandleByValArgument - When inlining a call site that has a byval argument,
/// we have to make the implicit memcpy explicit by adding it.
static Value *HandleByValArgument(Value *Arg, Instruction *TheCall,
                                  const Function *CalledFunc,
                                  InlineFunctionInfo &IFI,
                                  unsigned ByValAlignment) {
  PointerType *ArgTy = cast<PointerType>(Arg->getType());
  Type *AggTy = ArgTy->getElementType();

  // If the called function is readonly, then it could not mutate the caller's
  // copy of the byval'd memory.  In this case, it is safe to elide the copy and
  // temporary.
  if (CalledFunc->onlyReadsMemory()) {
    // If the byval argument has a specified alignment that is greater than the
    // passed in pointer, then we either have to round up the input pointer or
    // give up on this transformation.
    if (ByValAlignment <= 1)  // 0 = unspecified, 1 = no particular alignment.
      return Arg;

    // If the pointer is already known to be sufficiently aligned, or if we can
    // round it up to a larger alignment, then we don't need a temporary.
    if (getOrEnforceKnownAlignment(Arg, ByValAlignment,
                                   IFI.DL, IFI.AT, TheCall) >= ByValAlignment)
      return Arg;
    
    // Otherwise, we have to make a memcpy to get a safe alignment.  This is bad
    // for code quality, but rarely happens and is required for correctness.
  }

  // Create the alloca.  If we have DataLayout, use nice alignment.
  unsigned Align = 1;
  if (IFI.DL)
    Align = IFI.DL->getPrefTypeAlignment(AggTy);
  
  // If the byval had an alignment specified, we *must* use at least that
  // alignment, as it is required by the byval argument (and uses of the
  // pointer inside the callee).
  Align = std::max(Align, ByValAlignment);
  
  Function *Caller = TheCall->getParent()->getParent(); 
  
  Value *NewAlloca = new AllocaInst(AggTy, nullptr, Align, Arg->getName(), 
                                    &*Caller->begin()->begin());
  IFI.StaticAllocas.push_back(cast<AllocaInst>(NewAlloca));
  
  // Uses of the argument in the function should use our new alloca
  // instead.
  return NewAlloca;
}

// isUsedByLifetimeMarker - Check whether this Value is used by a lifetime
// intrinsic.
static bool isUsedByLifetimeMarker(Value *V) {
  for (User *U : V->users()) {
    if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(U)) {
      switch (II->getIntrinsicID()) {
      default: break;
      case Intrinsic::lifetime_start:
      case Intrinsic::lifetime_end:
        return true;
      }
    }
  }
  return false;
}

// hasLifetimeMarkers - Check whether the given alloca already has
// lifetime.start or lifetime.end intrinsics.
static bool hasLifetimeMarkers(AllocaInst *AI) {
  Type *Ty = AI->getType();
  Type *Int8PtrTy = Type::getInt8PtrTy(Ty->getContext(),
                                       Ty->getPointerAddressSpace());
  if (Ty == Int8PtrTy)
    return isUsedByLifetimeMarker(AI);

  // Do a scan to find all the casts to i8*.
  for (User *U : AI->users()) {
    if (U->getType() != Int8PtrTy) continue;
    if (U->stripPointerCasts() != AI) continue;
    if (isUsedByLifetimeMarker(U))
      return true;
  }
  return false;
}

/// updateInlinedAtInfo - Helper function used by fixupLineNumbers to
/// recursively update InlinedAtEntry of a DebugLoc.
static DebugLoc updateInlinedAtInfo(const DebugLoc &DL, 
                                    const DebugLoc &InlinedAtDL,
                                    LLVMContext &Ctx) {
  if (MDNode *IA = DL.getInlinedAt(Ctx)) {
    DebugLoc NewInlinedAtDL 
      = updateInlinedAtInfo(DebugLoc::getFromDILocation(IA), InlinedAtDL, Ctx);
    return DebugLoc::get(DL.getLine(), DL.getCol(), DL.getScope(Ctx),
                         NewInlinedAtDL.getAsMDNode(Ctx));
  }

  return DebugLoc::get(DL.getLine(), DL.getCol(), DL.getScope(Ctx),
                       InlinedAtDL.getAsMDNode(Ctx));
}

/// fixupLineNumbers - Update inlined instructions' line numbers to 
/// to encode location where these instructions are inlined.
static void fixupLineNumbers(Function *Fn, Function::iterator FI,
                             Instruction *TheCall) {
  DebugLoc TheCallDL = TheCall->getDebugLoc();
  if (TheCallDL.isUnknown())
    return;

  for (; FI != Fn->end(); ++FI) {
    for (BasicBlock::iterator BI = FI->begin(), BE = FI->end();
         BI != BE; ++BI) {
      DebugLoc DL = BI->getDebugLoc();
      if (DL.isUnknown()) {
        // If the inlined instruction has no line number, make it look as if it
        // originates from the call location. This is important for
        // ((__always_inline__, __nodebug__)) functions which must use caller
        // location for all instructions in their function body.
        BI->setDebugLoc(TheCallDL);
      } else {
        BI->setDebugLoc(updateInlinedAtInfo(DL, TheCallDL, BI->getContext()));
        if (DbgValueInst *DVI = dyn_cast<DbgValueInst>(BI)) {
          LLVMContext &Ctx = BI->getContext();
          MDNode *InlinedAt = BI->getDebugLoc().getInlinedAt(Ctx);
          DVI->setOperand(2, createInlinedVariable(DVI->getVariable(), 
                                                   InlinedAt, Ctx));
        }
      }
    }
  }
}

/// InlineFunction - This function inlines the called function into the basic
/// block of the caller.  This returns false if it is not possible to inline
/// this call.  The program is still in a well defined state if this occurs
/// though.
///
/// Note that this only does one level of inlining.  For example, if the
/// instruction 'call B' is inlined, and 'B' calls 'C', then the call to 'C' now
/// exists in the instruction stream.  Similarly this will inline a recursive
/// function by one level.
bool llvm::InlineFunction(CallSite CS, InlineFunctionInfo &IFI,
                          bool InsertLifetime) {
  Instruction *TheCall = CS.getInstruction();
  assert(TheCall->getParent() && TheCall->getParent()->getParent() &&
         "Instruction not in function!");

  // If IFI has any state in it, zap it before we fill it in.
  IFI.reset();
  
  const Function *CalledFunc = CS.getCalledFunction();
  if (!CalledFunc ||              // Can't inline external function or indirect
      CalledFunc->isDeclaration() || // call, or call to a vararg function!
      CalledFunc->getFunctionType()->isVarArg()) return false;

  // If the call to the callee cannot throw, set the 'nounwind' flag on any
  // calls that we inline.
  bool MarkNoUnwind = CS.doesNotThrow();

  BasicBlock *OrigBB = TheCall->getParent();
  Function *Caller = OrigBB->getParent();

  // GC poses two hazards to inlining, which only occur when the callee has GC:
  //  1. If the caller has no GC, then the callee's GC must be propagated to the
  //     caller.
  //  2. If the caller has a differing GC, it is invalid to inline.
  if (CalledFunc->hasGC()) {
    if (!Caller->hasGC())
      Caller->setGC(CalledFunc->getGC());
    else if (CalledFunc->getGC() != Caller->getGC())
      return false;
  }

  // Get the personality function from the callee if it contains a landing pad.
  Value *CalleePersonality = nullptr;
  for (Function::const_iterator I = CalledFunc->begin(), E = CalledFunc->end();
       I != E; ++I)
    if (const InvokeInst *II = dyn_cast<InvokeInst>(I->getTerminator())) {
      const BasicBlock *BB = II->getUnwindDest();
      const LandingPadInst *LP = BB->getLandingPadInst();
      CalleePersonality = LP->getPersonalityFn();
      break;
    }

  // Find the personality function used by the landing pads of the caller. If it
  // exists, then check to see that it matches the personality function used in
  // the callee.
  if (CalleePersonality) {
    for (Function::const_iterator I = Caller->begin(), E = Caller->end();
         I != E; ++I)
      if (const InvokeInst *II = dyn_cast<InvokeInst>(I->getTerminator())) {
        const BasicBlock *BB = II->getUnwindDest();
        const LandingPadInst *LP = BB->getLandingPadInst();

        // If the personality functions match, then we can perform the
        // inlining. Otherwise, we can't inline.
        // TODO: This isn't 100% true. Some personality functions are proper
        //       supersets of others and can be used in place of the other.
        if (LP->getPersonalityFn() != CalleePersonality)
          return false;

        break;
      }
  }

  // Get an iterator to the last basic block in the function, which will have
  // the new function inlined after it.
  Function::iterator LastBlock = &Caller->back();

  // Make sure to capture all of the return instructions from the cloned
  // function.
  SmallVector<ReturnInst*, 8> Returns;
  ClonedCodeInfo InlinedFunctionInfo;
  Function::iterator FirstNewBlock;

  { // Scope to destroy VMap after cloning.
    ValueToValueMapTy VMap;
    // Keep a list of pair (dst, src) to emit byval initializations.
    SmallVector<std::pair<Value*, Value*>, 4> ByValInit;

    assert(CalledFunc->arg_size() == CS.arg_size() &&
           "No varargs calls can be inlined!");

    // Calculate the vector of arguments to pass into the function cloner, which
    // matches up the formal to the actual argument values.
    CallSite::arg_iterator AI = CS.arg_begin();
    unsigned ArgNo = 0;
    for (Function::const_arg_iterator I = CalledFunc->arg_begin(),
         E = CalledFunc->arg_end(); I != E; ++I, ++AI, ++ArgNo) {
      Value *ActualArg = *AI;

      // When byval arguments actually inlined, we need to make the copy implied
      // by them explicit.  However, we don't do this if the callee is readonly
      // or readnone, because the copy would be unneeded: the callee doesn't
      // modify the struct.
      if (CS.isByValArgument(ArgNo)) {
        ActualArg = HandleByValArgument(ActualArg, TheCall, CalledFunc, IFI,
                                        CalledFunc->getParamAlignment(ArgNo+1));
        if (ActualArg != *AI)
          ByValInit.push_back(std::make_pair(ActualArg, (Value*) *AI));
      }

      VMap[I] = ActualArg;
    }

    // Add alignment assumptions if necessary. We do this before the inlined
    // instructions are actually cloned into the caller so that we can easily
    // check what will be known at the start of the inlined code.
    AddAlignmentAssumptions(CS, IFI);

    // We want the inliner to prune the code as it copies.  We would LOVE to
    // have no dead or constant instructions leftover after inlining occurs
    // (which can happen, e.g., because an argument was constant), but we'll be
    // happy with whatever the cloner can do.
    CloneAndPruneFunctionInto(Caller, CalledFunc, VMap, 
                              /*ModuleLevelChanges=*/false, Returns, ".i",
                              &InlinedFunctionInfo, IFI.DL, TheCall);

    // Remember the first block that is newly cloned over.
    FirstNewBlock = LastBlock; ++FirstNewBlock;