[APINotes] Support annotating C++ methods

This adds support for adding Clang attributes to C++ methods declared
within C++ records by using API Notes.

For instance:
```
Tags:
- Name: IntWrapper
  Methods:
  - Name: getIncremented
    Availability: none
```

This is the first instance of something within a C++ record being
annotated with API Notes, so it adds the necessary infra to make a C++
record an "API Notes context".

Notably this does not add support for nested C++ tags. That will be
added in a follow-up patch.

rdar://131387880

Author: egorzhdan

clang/include/clang/APINotes/APINotesReader.h

@@ -141,6 +141,16 @@ class APINotesReader {
                                                  ObjCSelectorRef Selector,
                                                  bool IsInstanceMethod);
 
+  /// Look for information regarding the given C++ method in the given C++ tag
+  /// context.
+  ///
+  /// \param CtxID The ID that references the parent context, i.e. a C++ tag.
+  /// \param Name The name of the C++ method we're looking for.
+  ///
+  /// \returns Information about the method, if known.
+  VersionedInfo<CXXMethodInfo> lookupCXXMethod(ContextID CtxID,
+                                               llvm::StringRef Name);
+
   /// Look for information regarding the given global variable.
   ///
   /// \param Name The name of the global variable.
@@ -166,6 +176,17 @@ class APINotesReader {
   /// \returns information about the enumerator, if known.
   VersionedInfo<EnumConstantInfo> lookupEnumConstant(llvm::StringRef Name);
 
+  /// Look for the context ID of the given C++ tag.
+  ///
+  /// \param Name The name of the tag we're looking for.
+  /// \param ParentCtx The context in which this tag is declared, e.g. a C++
+  /// namespace.
+  ///
+  /// \returns The ID, if known.
+  std::optional<ContextID>
+  lookupTagID(llvm::StringRef Name,
+              std::optional<Context> ParentCtx = std::nullopt);
+
   /// Look for information regarding the given tag
   /// (struct/union/enum/C++ class).
   ///

clang/include/clang/APINotes/APINotesWriter.h

@@ -78,6 +78,14 @@ class APINotesWriter {
                      bool IsInstanceMethod, const ObjCMethodInfo &Info,
                      llvm::VersionTuple SwiftVersion);
 
+  /// Add information about a specific C++ method.
+  ///
+  /// \param CtxID The context in which this method resides, i.e. a C++ tag.
+  /// \param Name The name of the method.
+  /// \param Info Information about this method.
+  void addCXXMethod(ContextID CtxID, llvm::StringRef Name,
+                    const CXXMethodInfo &Info, llvm::VersionTuple SwiftVersion);
+
   /// Add information about a global variable.
   ///
   /// \param Name The name of this global variable.

clang/include/clang/APINotes/Types.h

@@ -656,6 +656,12 @@ class GlobalFunctionInfo : public FunctionInfo {
   GlobalFunctionInfo() {}
 };
 
+/// Describes API notes data for a C++ method.
+class CXXMethodInfo : public FunctionInfo {
+public:
+  CXXMethodInfo() {}
+};
+
 /// Describes API notes data for an enumerator.
 class EnumConstantInfo : public CommonEntityInfo {
 public:
@@ -789,6 +795,7 @@ enum class ContextKind : uint8_t {
   ObjCClass = 0,
   ObjCProtocol = 1,
   Namespace = 2,
+  Tag = 3,
 };
 
 struct Context {

clang/lib/APINotes/APINotesFormat.h

@@ -63,6 +63,10 @@ enum BlockID {
   /// about the method.
   OBJC_METHOD_BLOCK_ID,
 
+  /// The C++ method data block, which maps C++ (context id, method name) pairs
+  /// to information about the method.
+  CXX_METHOD_BLOCK_ID,
+
   /// The Objective-C selector data block, which maps Objective-C
   /// selector names (# of pieces, identifier IDs) to the selector ID
   /// used in other tables.
@@ -181,6 +185,20 @@ using ObjCMethodDataLayout =
                          >;
 } // namespace objc_method_block
 
+namespace cxx_method_block {
+enum {
+  CXX_METHOD_DATA = 1,
+};
+
+using CXXMethodDataLayout =
+    llvm::BCRecordLayout<CXX_METHOD_DATA, // record ID
+                         llvm::BCVBR<16>, // table offset within the blob (see
+                                          // below)
+                         llvm::BCBlob     // map from C++ (context id, name)
+                                          // tuples to C++ method information
+                         >;
+} // namespace cxx_method_block
+
 namespace objc_selector_block {
 enum {
   OBJC_SELECTOR_DATA = 1,
@@ -269,6 +287,17 @@ struct ContextTableKey {
       : parentContextID(parentContextID), contextKind(contextKind),
         contextID(contextID) {}
 
+  ContextTableKey(std::optional<ContextID> ParentContextID, ContextKind Kind,
+                  uint32_t ContextID)
+      : parentContextID(ParentContextID ? ParentContextID->Value : -1),
+        contextKind(static_cast<uint8_t>(Kind)), contextID(ContextID) {}
+
+  ContextTableKey(std::optional<Context> ParentContext, ContextKind Kind,
+                  uint32_t ContextID)
+      : ContextTableKey(ParentContext ? std::make_optional(ParentContext->id)
+                                      : std::nullopt,
+                        Kind, ContextID) {}
+
   llvm::hash_code hashValue() const {
     return llvm::hash_value(
         std::tuple{parentContextID, contextKind, contextID});

clang/lib/APINotes/APINotesReader.cpp

@@ -473,6 +473,29 @@ class GlobalFunctionTableInfo
   }
 };
 
+/// Used to deserialize the on-disk C++ method table.
+class CXXMethodTableInfo
+    : public VersionedTableInfo<CXXMethodTableInfo, SingleDeclTableKey,
+                                CXXMethodInfo> {
+public:
+  static internal_key_type ReadKey(const uint8_t *Data, unsigned Length) {
+    auto CtxID = endian::readNext<uint32_t, llvm::endianness::little>(Data);
+    auto NameID = endian::readNext<uint32_t, llvm::endianness::little>(Data);
+    return {CtxID, NameID};
+  }
+
+  hash_value_type ComputeHash(internal_key_type Key) {
+    return static_cast<size_t>(Key.hashValue());
+  }
+
+  static CXXMethodInfo readUnversioned(internal_key_type Key,
+                                       const uint8_t *&Data) {
+    CXXMethodInfo Info;
+    ReadFunctionInfo(Data, Info);
+    return Info;
+  }
+};
+
 /// Used to deserialize the on-disk enumerator table.
 class EnumConstantTableInfo
     : public VersionedTableInfo<EnumConstantTableInfo, uint32_t,
@@ -630,6 +653,12 @@ class APINotesReader::Implementation {
   /// The Objective-C method table.
   std::unique_ptr<SerializedObjCMethodTable> ObjCMethodTable;
 
+  using SerializedCXXMethodTable =
+      llvm::OnDiskIterableChainedHashTable<CXXMethodTableInfo>;
+
+  /// The C++ method table.
+  std::unique_ptr<SerializedCXXMethodTable> CXXMethodTable;
+
   using SerializedObjCSelectorTable =
       llvm::OnDiskIterableChainedHashTable<ObjCSelectorTableInfo>;
 
@@ -683,6 +712,8 @@ class APINotesReader::Implementation {
                              llvm::SmallVectorImpl<uint64_t> &Scratch);
   bool readObjCMethodBlock(llvm::BitstreamCursor &Cursor,
                            llvm::SmallVectorImpl<uint64_t> &Scratch);
+  bool readCXXMethodBlock(llvm::BitstreamCursor &Cursor,
+                          llvm::SmallVectorImpl<uint64_t> &Scratch);
   bool readObjCSelectorBlock(llvm::BitstreamCursor &Cursor,
                              llvm::SmallVectorImpl<uint64_t> &Scratch);
   bool readGlobalVariableBlock(llvm::BitstreamCursor &Cursor,
@@ -1140,6 +1171,81 @@ bool APINotesReader::Implementation::readObjCMethodBlock(
   return false;
 }
 
+bool APINotesReader::Implementation::readCXXMethodBlock(
+    llvm::BitstreamCursor &Cursor, llvm::SmallVectorImpl<uint64_t> &Scratch) {
+  if (Cursor.EnterSubBlock(CXX_METHOD_BLOCK_ID))
+    return true;
+
+  llvm::Expected<llvm::BitstreamEntry> MaybeNext = Cursor.advance();
+  if (!MaybeNext) {
+    // FIXME this drops the error on the floor.
+    consumeError(MaybeNext.takeError());
+    return false;
+  }
+  llvm::BitstreamEntry Next = MaybeNext.get();
+  while (Next.Kind != llvm::BitstreamEntry::EndBlock) {
+    if (Next.Kind == llvm::BitstreamEntry::Error)
+      return true;
+
+    if (Next.Kind == llvm::BitstreamEntry::SubBlock) {
+      // Unknown sub-block, possibly for use by a future version of the
+      // API notes format.
+      if (Cursor.SkipBlock())
+        return true;
+
+      MaybeNext = Cursor.advance();
+      if (!MaybeNext) {
+        // FIXME this drops the error on the floor.
+        consumeError(MaybeNext.takeError());
+        return false;
+      }
+      Next = MaybeNext.get();
+      continue;
+    }
+
+    Scratch.clear();
+    llvm::StringRef BlobData;
+    llvm::Expected<unsigned> MaybeKind =
+        Cursor.readRecord(Next.ID, Scratch, &BlobData);
+    if (!MaybeKind) {
+      // FIXME this drops the error on the floor.
+      consumeError(MaybeKind.takeError());
+      return false;
+    }
+    unsigned Kind = MaybeKind.get();
+    switch (Kind) {
+    case cxx_method_block::CXX_METHOD_DATA: {
+      // Already saw C++ method table.
+      if (CXXMethodTable)
+        return true;
+
+      uint32_t tableOffset;
+      cxx_method_block::CXXMethodDataLayout::readRecord(Scratch, tableOffset);
+      auto base = reinterpret_cast<const uint8_t *>(BlobData.data());
+
+      CXXMethodTable.reset(SerializedCXXMethodTable::Create(
+          base + tableOffset, base + sizeof(uint32_t), base));
+      break;
+    }
+
+    default:
+      // Unknown record, possibly for use by a future version of the
+      // module format.
+      break;
+    }
+
+    MaybeNext = Cursor.advance();
+    if (!MaybeNext) {
+      // FIXME this drops the error on the floor.
+      consumeError(MaybeNext.takeError());
+      return false;
+    }
+    Next = MaybeNext.get();
+  }
+
+  return false;
+}
+
 bool APINotesReader::Implementation::readObjCSelectorBlock(
     llvm::BitstreamCursor &Cursor, llvm::SmallVectorImpl<uint64_t> &Scratch) {
   if (Cursor.EnterSubBlock(OBJC_SELECTOR_BLOCK_ID))
@@ -1692,6 +1798,14 @@ APINotesReader::APINotesReader(llvm::MemoryBuffer *InputBuffer,
       }
       break;
 
+    case CXX_METHOD_BLOCK_ID:
+      if (!HasValidControlBlock ||
+          Implementation->readCXXMethodBlock(Cursor, Scratch)) {
+        Failed = true;
+        return;
+      }
+      break;
+
     case OBJC_SELECTOR_BLOCK_ID:
       if (!HasValidControlBlock ||
           Implementation->readObjCSelectorBlock(Cursor, Scratch)) {
@@ -1911,6 +2025,23 @@ auto APINotesReader::lookupObjCMethod(ContextID CtxID, ObjCSelectorRef Selector,
   return {Implementation->SwiftVersion, *Known};
 }
 
+auto APINotesReader::lookupCXXMethod(ContextID CtxID, llvm::StringRef Name)
+    -> VersionedInfo<CXXMethodInfo> {
+  if (!Implementation->CXXMethodTable)
+    return std::nullopt;
+
+  std::optional<IdentifierID> NameID = Implementation->getIdentifier(Name);
+  if (!NameID)
+    return std::nullopt;
+
+  auto Known = Implementation->CXXMethodTable->find(
+      SingleDeclTableKey(CtxID.Value, *NameID));
+  if (Known == Implementation->CXXMethodTable->end())
+    return std::nullopt;
+
+  return {Implementation->SwiftVersion, *Known};
+}
+
 auto APINotesReader::lookupGlobalVariable(llvm::StringRef Name,
                                           std::optional<Context> Ctx)
     -> VersionedInfo<GlobalVariableInfo> {
@@ -1965,6 +2096,24 @@ auto APINotesReader::lookupEnumConstant(llvm::StringRef Name)
   return {Implementation->SwiftVersion, *Known};
 }
 
+auto APINotesReader::lookupTagID(llvm::StringRef Name,
+                                 std::optional<Context> ParentCtx)
+    -> std::optional<ContextID> {
+  if (!Implementation->ContextIDTable)
+    return std::nullopt;
+
+  std::optional<IdentifierID> TagID = Implementation->getIdentifier(Name);
+  if (!TagID)
+    return std::nullopt;
+
+  auto KnownID = Implementation->ContextIDTable->find(
+      ContextTableKey(ParentCtx, ContextKind::Tag, *TagID));
+  if (KnownID == Implementation->ContextIDTable->end())
+    return std::nullopt;
+
+  return ContextID(*KnownID);
+}
+
 auto APINotesReader::lookupTag(llvm::StringRef Name, std::optional<Context> Ctx)
     -> VersionedInfo<TagInfo> {
   if (!Implementation->TagTable)