CodeGenTBAA.cpp

Go to the documentation of this file.

//===-- CodeGenTBAA.cpp - TBAA information for LLVM CodeGen ---------------===//
//
// 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
//
//===----------------------------------------------------------------------===//
//
// This is the code that manages TBAA information and defines the TBAA policy
// for the optimizer to use. Relevant standards text includes:
//
//   C99 6.5p7
//   C++ [basic.lval] (p10 in n3126, p15 in some earlier versions)
//
//===----------------------------------------------------------------------===//
 
#include "CodeGenTBAA.h"
#include "ABIInfoImpl.h"
#include "CGCXXABI.h"
#include "CGRecordLayout.h"
#include "CodeGenTypes.h"
#include "clang/AST/ASTContext.h"
#include "clang/AST/Attr.h"
#include "clang/AST/Mangle.h"
#include "clang/AST/RecordLayout.h"
#include "clang/Basic/CodeGenOptions.h"
#include "clang/Basic/TargetInfo.h"
#include "clang/CodeGenUtils/RecordLayoutUtils.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Metadata.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/Type.h"
#include "llvm/Support/Debug.h"
using namespace clang;
using namespace CodeGen;
 
CodeGenTBAA::CodeGenTBAA(ASTContext &Ctx, CodeGenTypes &CGTypes,
                         llvm::Module &M, const CodeGenOptions &CGO,
                         const LangOptions &Features)
    : Context(Ctx), CGTypes(CGTypes), Module(M), CodeGenOpts(CGO),
      Features(Features),
      MangleCtx(ItaniumMangleContext::create(Ctx, Ctx.getDiagnostics())),
      MDHelper(M.getContext()), Root(nullptr), Char(nullptr) {}
 
CodeGenTBAA::~CodeGenTBAA() {
}
 
llvm::MDNode *CodeGenTBAA::getRoot() {
  // Define the root of the tree. This identifies the tree, so that
  // if our LLVM IR is linked with LLVM IR from a different front-end
  // (or a different version of this front-end), their TBAA trees will
  // remain distinct, and the optimizer will treat them conservatively.
  if (!Root) {
    if (Features.CPlusPlus)
      Root = MDHelper.createTBAARoot("Simple C++ TBAA");
    else
      Root = MDHelper.createTBAARoot("Simple C/C++ TBAA");
  }
 
  return Root;
}
 
llvm::MDNode *CodeGenTBAA::createScalarTypeNode(StringRef Name,
                                                llvm::MDNode *Parent,
                                                uint64_t Size) {
  if (CodeGenOpts.NewStructPathTBAA) {
    llvm::Metadata *Id = MDHelper.createString(Name);
    return MDHelper.createTBAATypeNode(Parent, Size, Id);
  }
  return MDHelper.createTBAAScalarTypeNode(Name, Parent);
}
 
llvm::MDNode *CodeGenTBAA::getChar() {
  // Define the root of the tree for user-accessible memory. C and C++
  // give special powers to char and certain similar types. However,
  // these special powers only cover user-accessible memory, and doesn't
  // include things like vtables.
  if (!Char)
    Char = createScalarTypeNode("omnipotent char", getRoot(), /* Size= */ 1);
 
  return Char;
}
 
llvm::MDNode *CodeGenTBAA::getAnyPtr(unsigned PtrDepth) {
  assert(PtrDepth >= 1 && "Pointer must have some depth");
 
  // Populate at least PtrDepth elements in AnyPtrs. These are the type nodes
  // for "any" pointers of increasing pointer depth, and are organized in the
  // hierarchy: any pointer <- any p2 pointer <- any p3 pointer <- ...
  //
  // Note that AnyPtrs[Idx] is actually the node for pointer depth (Idx+1),
  // since there is no node for pointer depth 0.
  //
  // These "any" pointer type nodes are used in pointer TBAA. The type node of
  // a concrete pointer type has the "any" pointer type node of appropriate
  // pointer depth as its parent. The "any" pointer type nodes are also used
  // directly for accesses to void pointers, or to specific pointers that we
  // conservatively do not distinguish in pointer TBAA (e.g. pointers to
  // members). Essentially, this establishes that e.g. void** can alias with
  // any type that can unify with T**, ignoring things like qualifiers. Here, T
  // is a variable that represents an arbitrary type, including pointer types.
  // As such, each depth is naturally a subtype of the previous depth, and thus
  // transitively of all previous depths.
  if (AnyPtrs.size() < PtrDepth) {
    AnyPtrs.reserve(PtrDepth);
    auto Size = Module.getDataLayout().getPointerSize();
    // Populate first element.
    if (AnyPtrs.empty())
      AnyPtrs.push_back(createScalarTypeNode("any pointer", getChar(), Size));
    // Populate further elements.
    for (size_t Idx = AnyPtrs.size(); Idx < PtrDepth; ++Idx) {
      auto Name = ("any p" + llvm::Twine(Idx + 1) + " pointer").str();
      AnyPtrs.push_back(createScalarTypeNode(Name, AnyPtrs[Idx - 1], Size));
    }
  }
 
  return AnyPtrs[PtrDepth - 1];
}
 
static bool TypeHasMayAlias(QualType QTy) {
  // Tagged types have declarations, and therefore may have attributes.
  if (auto *TD = QTy->getAsTagDecl())
    if (TD->hasAttr<MayAliasAttr>())
      return true;
 
  // Also look for may_alias as a declaration attribute on a typedef.
  // FIXME: We should follow GCC and model may_alias as a type attribute
  // rather than as a declaration attribute.
  while (auto *TT = QTy->getAs<TypedefType>()) {
    if (TT->getDecl()->hasAttr<MayAliasAttr>())
      return true;
    QTy = TT->desugar();
  }
 
  // Also consider an array type as may_alias when its element type (at
  // any level) is marked as such.
  if (auto *ArrayTy = QTy->getAsArrayTypeUnsafe())
    if (TypeHasMayAlias(ArrayTy->getElementType()))
      return true;
 
  return false;
}
 
/// Check if the given type is a valid base type to be used in access tags.
static bool isValidBaseType(QualType QTy) {
  if (const auto *RD = QTy->getAsRecordDecl()) {
    // Incomplete types are not valid base access types.
    if (!RD->isCompleteDefinition())
      return false;
    if (RD->hasFlexibleArrayMember())
      return false;
    // RD can be struct, union, class, interface or enum.
    // For now, we only handle struct and class.
    if (RD->isStruct() || RD->isClass())
      return true;
  }
  return false;
}
 
llvm::MDNode *CodeGenTBAA::getTypeInfoHelper(const Type *Ty) {
  uint64_t Size = Context.getTypeSizeInChars(Ty).getQuantity();
 
  // Handle builtin types.
  if (const BuiltinType *BTy = dyn_cast<BuiltinType>(Ty)) {
    switch (BTy->getKind()) {
    // Character types are special and can alias anything.
    // In C++, this technically only includes "char" and "unsigned char",
    // and not "signed char". In C, it includes all three. For now,
    // the risk of exploiting this detail in C++ seems likely to outweigh
    // the benefit.
    case BuiltinType::Char_U:
    case BuiltinType::Char_S:
    case BuiltinType::UChar:
    case BuiltinType::SChar:
      return getChar();
 
    // Unsigned types can alias their corresponding signed types.
    case BuiltinType::UShort:
      return getTypeInfo(Context.ShortTy);
    case BuiltinType::UInt:
      return getTypeInfo(Context.IntTy);
    case BuiltinType::ULong:
      return getTypeInfo(Context.LongTy);
    case BuiltinType::ULongLong:
      return getTypeInfo(Context.LongLongTy);
    case BuiltinType::UInt128:
      return getTypeInfo(Context.Int128Ty);
 
    case BuiltinType::UShortFract:
      return getTypeInfo(Context.ShortFractTy);
    case BuiltinType::UFract:
      return getTypeInfo(Context.FractTy);
    case BuiltinType::ULongFract:
      return getTypeInfo(Context.LongFractTy);
 
    case BuiltinType::SatUShortFract:
      return getTypeInfo(Context.SatShortFractTy);
    case BuiltinType::SatUFract:
      return getTypeInfo(Context.SatFractTy);
    case BuiltinType::SatULongFract:
      return getTypeInfo(Context.SatLongFractTy);
 
    case BuiltinType::UShortAccum:
      return getTypeInfo(Context.ShortAccumTy);
    case BuiltinType::UAccum:
      return getTypeInfo(Context.AccumTy);
    case BuiltinType::ULongAccum:
      return getTypeInfo(Context.LongAccumTy);
 
    case BuiltinType::SatUShortAccum:
      return getTypeInfo(Context.SatShortAccumTy);
    case BuiltinType::SatUAccum:
      return getTypeInfo(Context.SatAccumTy);
    case BuiltinType::SatULongAccum:
      return getTypeInfo(Context.SatLongAccumTy);
 
    // Treat all other builtin types as distinct types. This includes
    // treating wchar_t, char16_t, and char32_t as distinct from their
    // "underlying types".
    default:
      return createScalarTypeNode(BTy->getName(Features), getChar(), Size);
    }
  }
 
  // C++1z [basic.lval]p10: "If a program attempts to access the stored value of
  // an object through a glvalue of other than one of the following types the
  // behavior is undefined: [...] a char, unsigned char, or std::byte type."
  if (Ty->isStdByteType())
    return getChar();
 
  // Handle pointers and references.
  //
  // C has a very strict rule for pointer aliasing. C23 6.7.6.1p2:
  //     For two pointer types to be compatible, both shall be identically
  //     qualified and both shall be pointers to compatible types.
  //
  // This rule is impractically strict; we want to at least ignore CVR
  // qualifiers. Distinguishing by CVR qualifiers would make it UB to
  // e.g. cast a `char **` to `const char * const *` and dereference it,
  // which is too common and useful to invalidate. C++'s similar types
  // rule permits qualifier differences in these nested positions; in fact,
  // C++ even allows that cast as an implicit conversion.
  //
  // Other qualifiers could theoretically be distinguished, especially if
  // they involve a significant representation difference.  We don't
  // currently do so, however.
  if (Ty->isPointerType() || Ty->isReferenceType()) {
    if (!CodeGenOpts.PointerTBAA)
      return getAnyPtr();
    // C++ [basic.lval]p11 permits objects to accessed through an l-value of
    // similar type. Two types are similar under C++ [conv.qual]p2 if the
    // decomposition of the types into pointers, member pointers, and arrays has
    // the same structure when ignoring cv-qualifiers at each level of the
    // decomposition. Meanwhile, C makes T(*)[] and T(*)[N] compatible, which
    // would really complicate any attempt to distinguish pointers to arrays by
    // their bounds. It's simpler, and much easier to explain to users, to
    // simply treat all pointers to arrays as pointers to their element type for
    // aliasing purposes. So when creating a TBAA tag for a pointer type, we
    // recursively ignore both qualifiers and array types when decomposing the
    // pointee type. The only meaningful remaining structure is the number of
    // pointer types we encountered along the way, so we just produce the tag
    // "p<depth> <base type tag>". If we do find a member pointer type, for now
    // we just conservatively bail out with AnyPtr (below) rather than trying to
    // create a tag that honors the similar-type rules while still
    // distinguishing different kinds of member pointer.
    unsigned PtrDepth = 0;
    do {
      PtrDepth++;
      Ty = Ty->getPointeeType()->getBaseElementTypeUnsafe();
    } while (Ty->isPointerType());
 
    // While there are no special rules in the standards regarding void pointers
    // and strict aliasing, emitting distinct tags for void pointers break some
    // common idioms and there is no good alternative to re-write the code
    // without strict-aliasing violations.
    if (Ty->isVoidType())
      return getAnyPtr(PtrDepth);
 
    assert(!isa<VariableArrayType>(Ty));
    // When the underlying type is a builtin type, we compute the pointee type
    // string recursively, which is implicitly more forgiving than the standards
    // require.  Effectively, we are turning the question "are these types
    // compatible/similar" into "are accesses to these types allowed to alias".
    // In both C and C++, the latter question has special carve-outs for
    // signedness mismatches that only apply at the top level.  As a result, we
    // are allowing e.g. `int *` l-values to access `unsigned *` objects.
    SmallString<256> TyName;
    if (isa<BuiltinType>(Ty)) {
      llvm::MDNode *ScalarMD = getTypeInfoHelper(Ty);
      StringRef Name =
          cast<llvm::MDString>(
              ScalarMD->getOperand(CodeGenOpts.NewStructPathTBAA ? 2 : 0))
              ->getString();
      TyName = Name;
    } else {
      // Be conservative if the type isn't a RecordType. We are specifically
      // required to do this for member pointers until we implement the
      // similar-types rule.
      const auto *RT = Ty->getAsCanonical<RecordType>();
      if (!RT)
        return getAnyPtr(PtrDepth);
 
      // For unnamed structs or unions C's compatible types rule applies. Two
      // compatible types in different compilation units can have different
      // mangled names, meaning the metadata emitted below would incorrectly
      // mark them as no-alias. Use AnyPtr for such types in both C and C++, as
      // C and C++ types may be visible when doing LTO.
      //
      // Note that using AnyPtr is overly conservative. We could summarize the
      // members of the type, as per the C compatibility rule in the future.
      // This also covers anonymous structs and unions, which have a different
      // compatibility rule, but it doesn't matter because you can never have a
      // pointer to an anonymous struct or union.
      if (!RT->getDecl()->getDeclName())
        return getAnyPtr(PtrDepth);
 
      // For non-builtin types use the mangled name of the canonical type.
      llvm::raw_svector_ostream TyOut(TyName);
      MangleCtx->mangleCanonicalTypeName(QualType(Ty, 0), TyOut);
    }
 
    SmallString<256> OutName("p");
    OutName += std::to_string(PtrDepth);
    OutName += " ";
    OutName += TyName;
    return createScalarTypeNode(OutName, getAnyPtr(PtrDepth), Size);
  }
 
  // Accesses to arrays are accesses to objects of their element types.
  if (CodeGenOpts.NewStructPathTBAA && Ty->isArrayType())
    return getTypeInfo(cast<ArrayType>(Ty)->getElementType());
 
  // Accesses to matrix types are accesses to objects of their element types.
  if (const auto *MTy = dyn_cast<MatrixType>(Ty)) {
    assert(isa<ConstantMatrixType>(Ty) &&
           "only ConstantMatrixType should reach CodeGen");
    return getTypeInfo(MTy->getElementType());
  }
 
  // Accesses to complex types are accesses to objects of their element types.
  if (const auto *CTy = dyn_cast<ComplexType>(Ty))
    return getTypeInfo(CTy->getElementType());
 
  // Enum types are distinct types. In C++ they have "underlying types",
  // however they aren't related for TBAA.
  if (const EnumType *ETy = dyn_cast<EnumType>(Ty)) {
    const EnumDecl *ED = ETy->getDecl()->getDefinitionOrSelf();
    if (!Features.CPlusPlus)
      return getTypeInfo(ED->getIntegerType());
 
    // In C++ mode, types have linkage, so we can rely on the ODR and
    // on their mangled names, if they're external.
    // TODO: Is there a way to get a program-wide unique name for a
    // decl with local linkage or no linkage?
    if (!ED->isExternallyVisible())
      return getChar();
 
    SmallString<256> OutName;
    llvm::raw_svector_ostream Out(OutName);
    CGTypes.getCXXABI().getMangleContext().mangleCanonicalTypeName(
        QualType(ETy, 0), Out);
    return createScalarTypeNode(OutName, getChar(), Size);
  }
 
  if (const auto *EIT = dyn_cast<BitIntType>(Ty)) {
    SmallString<256> OutName;
    llvm::raw_svector_ostream Out(OutName);
    // Don't specify signed/unsigned since integer types can alias despite sign
    // differences.
    Out << "_BitInt(" << EIT->getNumBits() << ')';
    return createScalarTypeNode(OutName, getChar(), Size);
  }
 
  // For now, handle any other kind of type conservatively.
  return getChar();
}
 
llvm::MDNode *CodeGenTBAA::getTypeInfo(QualType QTy) {
  // At -O0 or relaxed aliasing, TBAA is not emitted for regular types (unless
  // we're running TypeSanitizer).
  if (!Features.Sanitize.has(SanitizerKind::Type) &&
      (CodeGenOpts.OptimizationLevel == 0 || CodeGenOpts.RelaxedAliasing))
    return nullptr;
 
  // If the type has the may_alias attribute (even on a typedef), it is
  // effectively in the general char alias class.
  if (TypeHasMayAlias(QTy))
    return getChar();
 
  // We need this function to not fall back to returning the "omnipotent char"
  // type node for aggregate and union types. Otherwise, any dereference of an
  // aggregate will result into the may-alias access descriptor, meaning all
  // subsequent accesses to direct and indirect members of that aggregate will
  // be considered may-alias too.
  // TODO: Combine getTypeInfo() and getValidBaseTypeInfo() into a single
  // function.
  if (isValidBaseType(QTy))
    return getValidBaseTypeInfo(QTy);
 
  const Type *Ty = Context.getCanonicalType(QTy).getTypePtr();
  if (llvm::MDNode *N = MetadataCache[Ty])
    return N;
 
  // Note that the following helper call is allowed to add new nodes to the
  // cache, which invalidates all its previously obtained iterators. So we
  // first generate the node for the type and then add that node to the cache.
  llvm::MDNode *TypeNode = getTypeInfoHelper(Ty);
  return MetadataCache[Ty] = TypeNode;
}
 
TBAAAccessInfo CodeGenTBAA::getAccessInfo(QualType AccessType) {
  // Pointee values may have incomplete types, but they shall never be
  // dereferenced.
  if (AccessType->isIncompleteType())
    return TBAAAccessInfo::getIncompleteInfo();
 
  if (TypeHasMayAlias(AccessType))
    return TBAAAccessInfo::getMayAliasInfo();
 
  uint64_t Size = Context.getTypeSizeInChars(AccessType).getQuantity();
  return TBAAAccessInfo(getTypeInfo(AccessType), Size);
}
 
TBAAAccessInfo CodeGenTBAA::getVTablePtrAccessInfo(llvm::Type *VTablePtrType) {
  const llvm::DataLayout &DL = Module.getDataLayout();
  unsigned Size = DL.getPointerTypeSize(VTablePtrType);
  return TBAAAccessInfo(createScalarTypeNode("vtable pointer", getRoot(), Size),
                        Size);
}
 
bool
CodeGenTBAA::CollectFields(uint64_t BaseOffset,
                           QualType QTy,
                           SmallVectorImpl<llvm::MDBuilder::TBAAStructField> &
                             Fields,
                           bool MayAlias) {
  /* Things not handled yet include: C++ base classes, bitfields, */
 
  if (const auto *TTy = QTy->getAsCanonical<RecordType>()) {
    if (TTy->isUnionType()) {
      uint64_t Size = Context.getTypeSizeInChars(QTy).getQuantity();
      llvm::MDNode *TBAAType = getChar();
      llvm::MDNode *TBAATag = getAccessTagInfo(TBAAAccessInfo(TBAAType, Size));
      Fields.push_back(
          llvm::MDBuilder::TBAAStructField(BaseOffset, Size, TBAATag));
      return true;
    }
    const RecordDecl *RD = TTy->getDecl()->getDefinition();
    if (RD->hasFlexibleArrayMember())
      return false;
 
    // TODO: Handle C++ base classes.
    if (const CXXRecordDecl *Decl = dyn_cast<CXXRecordDecl>(RD))
      if (!Decl->bases().empty())
        return false;
 
    const ASTRecordLayout &Layout = Context.getASTRecordLayout(RD);
    const CGRecordLayout &CGRL = CGTypes.getCGRecordLayout(RD);
 
    unsigned idx = 0;
    for (RecordDecl::field_iterator i = RD->field_begin(), e = RD->field_end();
         i != e; ++i, ++idx) {
      if (CodeGenUtils::isEmptyFieldForLayout(Context, *i))
        continue;
 
      uint64_t Offset =
          BaseOffset + Layout.getFieldOffset(idx) / Context.getCharWidth();
 
      // Create a single field for consecutive named bitfields using char as
      // base type.
      if ((*i)->isBitField()) {
        const CGBitFieldInfo &Info = CGRL.getBitFieldInfo(*i);
        // For big endian targets the first bitfield in the consecutive run is
        // at the most-significant end; see CGRecordLowering::setBitFieldInfo
        // for more information.
        bool IsBE = Context.getTargetInfo().isBigEndian();
        bool IsFirst = IsBE ? Info.StorageSize - (Info.Offset + Info.Size) == 0
                            : Info.Offset == 0;
        if (!IsFirst)
          continue;
        unsigned CurrentBitFieldSize = Info.StorageSize;
        uint64_t Size =
            llvm::divideCeil(CurrentBitFieldSize, Context.getCharWidth());
        llvm::MDNode *TBAAType = getChar();
        llvm::MDNode *TBAATag =
            getAccessTagInfo(TBAAAccessInfo(TBAAType, Size));
        Fields.push_back(
            llvm::MDBuilder::TBAAStructField(Offset, Size, TBAATag));
        continue;
      }
 
      QualType FieldQTy = i->getType();
      if (!CollectFields(Offset, FieldQTy, Fields,
                         MayAlias || TypeHasMayAlias(FieldQTy)))
        return false;
    }
    return true;
  }
 
  if (const auto *CTy = QTy->getAsCanonical<ComplexType>()) {
    QualType ElemQTy = CTy->getElementType();
    uint64_t Size = Context.getTypeSizeInChars(ElemQTy).getQuantity();
    llvm::MDNode *TBAAType = MayAlias ? getChar() : getTypeInfo(ElemQTy);
    llvm::MDNode *TBAATag = getAccessTagInfo(TBAAAccessInfo(TBAAType, Size));
    Fields.push_back(
        llvm::MDBuilder::TBAAStructField(BaseOffset, Size, TBAATag));
    Fields.push_back(
        llvm::MDBuilder::TBAAStructField(BaseOffset + Size, Size, TBAATag));
    return true;
  }
 
  /* Otherwise, treat whatever it is as a field. */
  uint64_t Offset = BaseOffset;
  uint64_t Size = Context.getTypeSizeInChars(QTy).getQuantity();
  llvm::MDNode *TBAAType = MayAlias ? getChar() : getTypeInfo(QTy);
  llvm::MDNode *TBAATag = getAccessTagInfo(TBAAAccessInfo(TBAAType, Size));
  Fields.push_back(llvm::MDBuilder::TBAAStructField(Offset, Size, TBAATag));
  return true;
}
 
llvm::MDNode *
CodeGenTBAA::getTBAAStructInfo(QualType QTy) {
  if (CodeGenOpts.OptimizationLevel == 0 || CodeGenOpts.RelaxedAliasing)
    return nullptr;
 
  const Type *Ty = Context.getCanonicalType(QTy).getTypePtr();
 
  if (llvm::MDNode *N = StructMetadataCache[Ty])
    return N;
 
  SmallVector<llvm::MDBuilder::TBAAStructField, 4> Fields;
  if (CollectFields(0, QTy, Fields, TypeHasMayAlias(QTy)))
    return MDHelper.createTBAAStructNode(Fields);
 
  // For now, handle any other kind of type conservatively.
  return StructMetadataCache[Ty] = nullptr;
}
 
llvm::MDNode *CodeGenTBAA::getBaseTypeInfoHelper(const Type *Ty) {
  if (auto *TTy = dyn_cast<RecordType>(Ty)) {
    const RecordDecl *RD = TTy->getDecl()->getDefinition();
    const ASTRecordLayout &Layout = Context.getASTRecordLayout(RD);
    using TBAAStructField = llvm::MDBuilder::TBAAStructField;
    SmallVector<TBAAStructField, 4> Fields;
    if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD)) {
      // Handle C++ base classes. Non-virtual bases can treated a kind of
      // field. Virtual bases are more complex and omitted, but avoid an
      // incomplete view for NewStructPathTBAA.
      if (CodeGenOpts.NewStructPathTBAA && CXXRD->getNumVBases() != 0)
        return nullptr;
      for (const CXXBaseSpecifier &B : CXXRD->bases()) {
        if (B.isVirtual())
          continue;
        QualType BaseQTy = B.getType();
        const CXXRecordDecl *BaseRD = BaseQTy->getAsCXXRecordDecl();
        if (BaseRD->isEmpty())
          continue;
        llvm::MDNode *TypeNode = isValidBaseType(BaseQTy)
                                     ? getValidBaseTypeInfo(BaseQTy)
                                     : getTypeInfo(BaseQTy);
        if (!TypeNode)
          return nullptr;
        uint64_t Offset = Layout.getBaseClassOffset(BaseRD).getQuantity();
        uint64_t Size =
            Context.getASTRecordLayout(BaseRD).getDataSize().getQuantity();
        Fields.push_back(
            llvm::MDBuilder::TBAAStructField(Offset, Size, TypeNode));
      }
      // The order in which base class subobjects are allocated is unspecified,
      // so may differ from declaration order. In particular, Itanium ABI will
      // allocate a primary base first.
      // Since we exclude empty subobjects, the objects are not overlapping and
      // their offsets are unique.
      llvm::sort(Fields,
                 [](const TBAAStructField &A, const TBAAStructField &B) {
                   return A.Offset < B.Offset;
                 });
    }
    for (FieldDecl *Field : RD->fields()) {
      if (Field->isZeroSize(Context) || Field->isUnnamedBitField())
        continue;
      QualType FieldQTy = Field->getType();
      llvm::MDNode *TypeNode = isValidBaseType(FieldQTy)
                                   ? getValidBaseTypeInfo(FieldQTy)
                                   : getTypeInfo(FieldQTy);
      if (!TypeNode)
        return nullptr;
 
      uint64_t BitOffset = Layout.getFieldOffset(Field->getFieldIndex());
      uint64_t Offset = Context.toCharUnitsFromBits(BitOffset).getQuantity();
      uint64_t Size = Context.getTypeSizeInChars(FieldQTy).getQuantity();
      Fields.push_back(llvm::MDBuilder::TBAAStructField(Offset, Size,
                                                        TypeNode));
    }
 
    SmallString<256> OutName;
    if (Features.CPlusPlus) {
      // Don't use the mangler for C code.
      llvm::raw_svector_ostream Out(OutName);
      CGTypes.getCXXABI().getMangleContext().mangleCanonicalTypeName(
          QualType(Ty, 0), Out);
    } else {
      OutName = RD->getName();
    }
 
    if (CodeGenOpts.NewStructPathTBAA) {
      llvm::MDNode *Parent = getChar();
      uint64_t Size = Context.getTypeSizeInChars(Ty).getQuantity();
      llvm::Metadata *Id = MDHelper.createString(OutName);
      return MDHelper.createTBAATypeNode(Parent, Size, Id, Fields);
    }
 
    // Create the struct type node with a vector of pairs (offset, type).
    SmallVector<std::pair<llvm::MDNode*, uint64_t>, 4> OffsetsAndTypes;
    for (const auto &Field : Fields)
        OffsetsAndTypes.push_back(std::make_pair(Field.Type, Field.Offset));
    return MDHelper.createTBAAStructTypeNode(OutName, OffsetsAndTypes);
  }
 
  return nullptr;
}
 
llvm::MDNode *CodeGenTBAA::getValidBaseTypeInfo(QualType QTy) {
  assert(isValidBaseType(QTy) && "Must be a valid base type");
 
  const Type *Ty = Context.getCanonicalType(QTy).getTypePtr();
 
  // nullptr is a valid value in the cache, so use find rather than []
  auto I = BaseTypeMetadataCache.find(Ty);
  if (I != BaseTypeMetadataCache.end())
    return I->second;
 
  // First calculate the metadata, before recomputing the insertion point, as
  // the helper can recursively call us.
  llvm::MDNode *TypeNode = getBaseTypeInfoHelper(Ty);
  [[maybe_unused]] auto inserted = BaseTypeMetadataCache.insert({Ty, TypeNode});
  assert(inserted.second && "BaseType metadata was already inserted");
 
  return TypeNode;
}
 
llvm::MDNode *CodeGenTBAA::getBaseTypeInfo(QualType QTy) {
  return isValidBaseType(QTy) ? getValidBaseTypeInfo(QTy) : nullptr;
}
 
llvm::MDNode *CodeGenTBAA::getAccessTagInfo(TBAAAccessInfo Info) {
  assert(!Info.isIncomplete() && "Access to an object of an incomplete type!");
 
  if (Info.isMayAlias())
    Info = TBAAAccessInfo(getChar(), Info.Size);
 
  if (!Info.AccessType)
    return nullptr;
 
  if (!CodeGenOpts.StructPathTBAA)
    Info = TBAAAccessInfo(Info.AccessType, Info.Size);
 
  llvm::MDNode *&N = AccessTagMetadataCache[Info];
  if (N)
    return N;
 
  if (!Info.BaseType) {
    Info.BaseType = Info.AccessType;
    assert(!Info.Offset && "Nonzero offset for an access with no base type!");
  }
  if (CodeGenOpts.NewStructPathTBAA) {
    return N = MDHelper.createTBAAAccessTag(Info.BaseType, Info.AccessType,
                                            Info.Offset, Info.Size);
  }
  return N = MDHelper.createTBAAStructTagNode(Info.BaseType, Info.AccessType,
                                              Info.Offset);
}
 
TBAAAccessInfo CodeGenTBAA::mergeTBAAInfoForCast(TBAAAccessInfo SourceInfo,
                                                 TBAAAccessInfo TargetInfo) {
  if (SourceInfo.isMayAlias() || TargetInfo.isMayAlias())
    return TBAAAccessInfo::getMayAliasInfo();
  return TargetInfo;
}
 
TBAAAccessInfo
CodeGenTBAA::mergeTBAAInfoForConditionalOperator(TBAAAccessInfo InfoA,
                                                 TBAAAccessInfo InfoB) {
  if (InfoA == InfoB)
    return InfoA;
 
  if (!InfoA || !InfoB)
    return TBAAAccessInfo();
 
  if (InfoA.isMayAlias() || InfoB.isMayAlias())
    return TBAAAccessInfo::getMayAliasInfo();
 
  // TODO: Implement the rest of the logic here. For example, two accesses
  // with same final access types result in an access to an object of that final
  // access type regardless of their base types.
  return TBAAAccessInfo::getMayAliasInfo();
}
 
TBAAAccessInfo
CodeGenTBAA::mergeTBAAInfoForMemoryTransfer(TBAAAccessInfo DestInfo,
                                            TBAAAccessInfo SrcInfo) {
  if (DestInfo == SrcInfo)
    return DestInfo;
 
  if (!DestInfo || !SrcInfo)
    return TBAAAccessInfo();
 
  if (DestInfo.isMayAlias() || SrcInfo.isMayAlias())
    return TBAAAccessInfo::getMayAliasInfo();
 
  // TODO: Implement the rest of the logic here. For example, two accesses
  // with same final access types result in an access to an object of that final
  // access type regardless of their base types.
  return TBAAAccessInfo::getMayAliasInfo();
}