C++ LoadInst::getMetadata方法代码示例

bool AMDGPUCodeGenPrepare::visitLoadInst(LoadInst &I) {
  if (!WidenLoads)
    return false;

  if ((I.getPointerAddressSpace() == AMDGPUASI.CONSTANT_ADDRESS ||
       I.getPointerAddressSpace() == AMDGPUASI.CONSTANT_ADDRESS_32BIT) &&
      canWidenScalarExtLoad(I)) {
    IRBuilder<> Builder(&I);
    Builder.SetCurrentDebugLocation(I.getDebugLoc());

    Type *I32Ty = Builder.getInt32Ty();
    Type *PT = PointerType::get(I32Ty, I.getPointerAddressSpace());
    Value *BitCast= Builder.CreateBitCast(I.getPointerOperand(), PT);
    LoadInst *WidenLoad = Builder.CreateLoad(BitCast);
    WidenLoad->copyMetadata(I);

    // If we have range metadata, we need to convert the type, and not make
    // assumptions about the high bits.
    if (auto *Range = WidenLoad->getMetadata(LLVMContext::MD_range)) {
      ConstantInt *Lower =
        mdconst::extract<ConstantInt>(Range->getOperand(0));

      if (Lower->getValue().isNullValue()) {
        WidenLoad->setMetadata(LLVMContext::MD_range, nullptr);
      } else {
        Metadata *LowAndHigh[] = {
          ConstantAsMetadata::get(ConstantInt::get(I32Ty, Lower->getValue().zext(32))),
          // Don't make assumptions about the high bits.
          ConstantAsMetadata::get(ConstantInt::get(I32Ty, 0))
        };

        WidenLoad->setMetadata(LLVMContext::MD_range,
                               MDNode::get(Mod->getContext(), LowAndHigh));
      }
    }

    int TySize = Mod->getDataLayout().getTypeSizeInBits(I.getType());
    Type *IntNTy = Builder.getIntNTy(TySize);
    Value *ValTrunc = Builder.CreateTrunc(WidenLoad, IntNTy);
    Value *ValOrig = Builder.CreateBitCast(ValTrunc, I.getType());
    I.replaceAllUsesWith(ValOrig);
    I.eraseFromParent();
    return true;
  }

  return false;
}

//.........这里部分代码省略.........
        // Store the Value* version of the indices in here, but declare it now
        // for reuse.
        std::vector<Value*> Ops;
        for (ScalarizeTable::iterator SI = ArgIndices.begin(),
               E = ArgIndices.end(); SI != E; ++SI) {
          Value *V = *AI;
          LoadInst *OrigLoad = OriginalLoads[*SI];
          if (!SI->empty()) {
            Ops.reserve(SI->size());
            const Type *ElTy = V->getType();
            for (IndicesVector::const_iterator II = SI->begin(),
                 IE = SI->end(); II != IE; ++II) {
              // Use i32 to index structs, and i64 for others (pointers/arrays).
              // This satisfies GEP constraints.
              const Type *IdxTy = (ElTy->isStructTy() ?
                    Type::getInt32Ty(F->getContext()) : 
                    Type::getInt64Ty(F->getContext()));
              Ops.push_back(ConstantInt::get(IdxTy, *II));
              // Keep track of the type we're currently indexing.
              ElTy = cast<CompositeType>(ElTy)->getTypeAtIndex(*II);
            }
            // And create a GEP to extract those indices.
            V = GetElementPtrInst::Create(V, Ops.begin(), Ops.end(),
                                          V->getName()+".idx", Call);
            Ops.clear();
            AA.copyValue(OrigLoad->getOperand(0), V);
          }
          // Since we're replacing a load make sure we take the alignment
          // of the previous load.
          LoadInst *newLoad = new LoadInst(V, V->getName()+".val", Call);
          newLoad->setAlignment(OrigLoad->getAlignment());
          // Transfer the TBAA info too.
          newLoad->setMetadata(LLVMContext::MD_tbaa,
                               OrigLoad->getMetadata(LLVMContext::MD_tbaa));
          Args.push_back(newLoad);
          AA.copyValue(OrigLoad, Args.back());
        }
      }

    if (ExtraArgHack)
      Args.push_back(Constant::getNullValue(Type::getInt32Ty(F->getContext())));

    // Push any varargs arguments on the list.
    for (; AI != CS.arg_end(); ++AI, ++ArgIndex) {
      Args.push_back(*AI);
      if (Attributes Attrs = CallPAL.getParamAttributes(ArgIndex))
        AttributesVec.push_back(AttributeWithIndex::get(Args.size(), Attrs));
    }

    // Add any function attributes.
    if (Attributes attrs = CallPAL.getFnAttributes())
      AttributesVec.push_back(AttributeWithIndex::get(~0, attrs));

    Instruction *New;
    if (InvokeInst *II = dyn_cast<InvokeInst>(Call)) {
      New = InvokeInst::Create(NF, II->getNormalDest(), II->getUnwindDest(),
                               Args.begin(), Args.end(), "", Call);
      cast<InvokeInst>(New)->setCallingConv(CS.getCallingConv());
      cast<InvokeInst>(New)->setAttributes(AttrListPtr::get(AttributesVec.begin(),
                                                          AttributesVec.end()));
    } else {
      New = CallInst::Create(NF, Args.begin(), Args.end(), "", Call);
      cast<CallInst>(New)->setCallingConv(CS.getCallingConv());
      cast<CallInst>(New)->setAttributes(AttrListPtr::get(AttributesVec.begin(),
                                                        AttributesVec.end()));
      if (cast<CallInst>(Call)->isTailCall())

/// Many allocas are only used within a single basic block.  If this is the
/// case, avoid traversing the CFG and inserting a lot of potentially useless
/// PHI nodes by just performing a single linear pass over the basic block
/// using the Alloca.
///
/// If we cannot promote this alloca (because it is read before it is written),
/// return false.  This is necessary in cases where, due to control flow, the
/// alloca is undefined only on some control flow paths.  e.g. code like
/// this is correct in LLVM IR:
///  // A is an alloca with no stores so far
///  for (...) {
///    int t = *A;
///    if (!first_iteration)
///      use(t);
///    *A = 42;
///  }
static bool promoteSingleBlockAlloca(AllocaInst *AI, const AllocaInfo &Info,
                                     LargeBlockInfo &LBI,
                                     const DataLayout &DL,
                                     DominatorTree &DT,
                                     AssumptionCache *AC) {
  // The trickiest case to handle is when we have large blocks. Because of this,
  // this code is optimized assuming that large blocks happen.  This does not
  // significantly pessimize the small block case.  This uses LargeBlockInfo to
  // make it efficient to get the index of various operations in the block.

  // Walk the use-def list of the alloca, getting the locations of all stores.
  using StoresByIndexTy = SmallVector<std::pair<unsigned, StoreInst *>, 64>;
  StoresByIndexTy StoresByIndex;

  for (User *U : AI->users())
    if (StoreInst *SI = dyn_cast<StoreInst>(U))
      StoresByIndex.push_back(std::make_pair(LBI.getInstructionIndex(SI), SI));

  // Sort the stores by their index, making it efficient to do a lookup with a
  // binary search.
  llvm::sort(StoresByIndex, less_first());

  // Walk all of the loads from this alloca, replacing them with the nearest
  // store above them, if any.
  for (auto UI = AI->user_begin(), E = AI->user_end(); UI != E;) {
    LoadInst *LI = dyn_cast<LoadInst>(*UI++);
    if (!LI)
      continue;

    unsigned LoadIdx = LBI.getInstructionIndex(LI);

    // Find the nearest store that has a lower index than this load.
    StoresByIndexTy::iterator I =
        std::lower_bound(StoresByIndex.begin(), StoresByIndex.end(),
                         std::make_pair(LoadIdx,
                                        static_cast<StoreInst *>(nullptr)),
                         less_first());
    if (I == StoresByIndex.begin()) {
      if (StoresByIndex.empty())
        // If there are no stores, the load takes the undef value.
        LI->replaceAllUsesWith(UndefValue::get(LI->getType()));
      else
        // There is no store before this load, bail out (load may be affected
        // by the following stores - see main comment).
        return false;
    } else {
      // Otherwise, there was a store before this load, the load takes its value.
      // Note, if the load was marked as nonnull we don't want to lose that
      // information when we erase it. So we preserve it with an assume.
      Value *ReplVal = std::prev(I)->second->getOperand(0);
      if (AC && LI->getMetadata(LLVMContext::MD_nonnull) &&
          !isKnownNonZero(ReplVal, DL, 0, AC, LI, &DT))
        addAssumeNonNull(AC, LI);

      // If the replacement value is the load, this must occur in unreachable
      // code.
      if (ReplVal == LI)
        ReplVal = UndefValue::get(LI->getType());

      LI->replaceAllUsesWith(ReplVal);
    }

    LI->eraseFromParent();
    LBI.deleteValue(LI);
  }

  // Remove the (now dead) stores and alloca.
  while (!AI->use_empty()) {
    StoreInst *SI = cast<StoreInst>(AI->user_back());
    // Record debuginfo for the store before removing it.
    for (DbgVariableIntrinsic *DII : Info.DbgDeclares) {
      DIBuilder DIB(*AI->getModule(), /*AllowUnresolved*/ false);
      ConvertDebugDeclareToDebugValue(DII, SI, DIB);
    }
    SI->eraseFromParent();
    LBI.deleteValue(SI);
  }

  AI->eraseFromParent();
  LBI.deleteValue(AI);

  // The alloca's debuginfo can be removed as well.
  for (DbgVariableIntrinsic *DII : Info.DbgDeclares) {
    DII->eraseFromParent();
//.........这里部分代码省略.........

/// Rewrite as many loads as possible given a single store.
///
/// When there is only a single store, we can use the domtree to trivially
/// replace all of the dominated loads with the stored value. Do so, and return
/// true if this has successfully promoted the alloca entirely. If this returns
/// false there were some loads which were not dominated by the single store
/// and thus must be phi-ed with undef. We fall back to the standard alloca
/// promotion algorithm in that case.
static bool rewriteSingleStoreAlloca(AllocaInst *AI, AllocaInfo &Info,
                                     LargeBlockInfo &LBI, const DataLayout &DL,
                                     DominatorTree &DT, AssumptionCache *AC) {
  StoreInst *OnlyStore = Info.OnlyStore;
  bool StoringGlobalVal = !isa<Instruction>(OnlyStore->getOperand(0));
  BasicBlock *StoreBB = OnlyStore->getParent();
  int StoreIndex = -1;

  // Clear out UsingBlocks.  We will reconstruct it here if needed.
  Info.UsingBlocks.clear();

  for (auto UI = AI->user_begin(), E = AI->user_end(); UI != E;) {
    Instruction *UserInst = cast<Instruction>(*UI++);
    if (!isa<LoadInst>(UserInst)) {
      assert(UserInst == OnlyStore && "Should only have load/stores");
      continue;
    }
    LoadInst *LI = cast<LoadInst>(UserInst);

    // Okay, if we have a load from the alloca, we want to replace it with the
    // only value stored to the alloca.  We can do this if the value is
    // dominated by the store.  If not, we use the rest of the mem2reg machinery
    // to insert the phi nodes as needed.
    if (!StoringGlobalVal) { // Non-instructions are always dominated.
      if (LI->getParent() == StoreBB) {
        // If we have a use that is in the same block as the store, compare the
        // indices of the two instructions to see which one came first.  If the
        // load came before the store, we can't handle it.
        if (StoreIndex == -1)
          StoreIndex = LBI.getInstructionIndex(OnlyStore);

        if (unsigned(StoreIndex) > LBI.getInstructionIndex(LI)) {
          // Can't handle this load, bail out.
          Info.UsingBlocks.push_back(StoreBB);
          continue;
        }
      } else if (LI->getParent() != StoreBB &&
                 !DT.dominates(StoreBB, LI->getParent())) {
        // If the load and store are in different blocks, use BB dominance to
        // check their relationships.  If the store doesn't dom the use, bail
        // out.
        Info.UsingBlocks.push_back(LI->getParent());
        continue;
      }
    }

    // Otherwise, we *can* safely rewrite this load.
    Value *ReplVal = OnlyStore->getOperand(0);
    // If the replacement value is the load, this must occur in unreachable
    // code.
    if (ReplVal == LI)
      ReplVal = UndefValue::get(LI->getType());

    // If the load was marked as nonnull we don't want to lose
    // that information when we erase this Load. So we preserve
    // it with an assume.
    if (AC && LI->getMetadata(LLVMContext::MD_nonnull) &&
        !isKnownNonZero(ReplVal, DL, 0, AC, LI, &DT))
      addAssumeNonNull(AC, LI);

    LI->replaceAllUsesWith(ReplVal);
    LI->eraseFromParent();
    LBI.deleteValue(LI);
  }

  // Finally, after the scan, check to see if the store is all that is left.
  if (!Info.UsingBlocks.empty())
    return false; // If not, we'll have to fall back for the remainder.

  // Record debuginfo for the store and remove the declaration's
  // debuginfo.
  for (DbgVariableIntrinsic *DII : Info.DbgDeclares) {
    DIBuilder DIB(*AI->getModule(), /*AllowUnresolved*/ false);
    ConvertDebugDeclareToDebugValue(DII, Info.OnlyStore, DIB);
    DII->eraseFromParent();
    LBI.deleteValue(DII);
  }
  // Remove the (now dead) store and alloca.
  Info.OnlyStore->eraseFromParent();
  LBI.deleteValue(Info.OnlyStore);

  AI->eraseFromParent();
  LBI.deleteValue(AI);
  return true;
}