Bitcode: Decouple block info block state from reader.

[pcc]

Summary

As proposed on llvm-dev:
http://lists.llvm.org/pipermail/llvm-dev/2016-October/106630.html

Move block info block state to a new class, BitstreamBlockInfo.
Clients may set the block info for a particular cursor with the
BitstreamCursor::setBlockInfo() method.

At this point BitstreamReader is not much more than a container for an
ArrayRef<uint8_t>, so remove it and replace all uses with direct uses
of memory buffers.

[pcc]

• Use implicit conversion to MemoryBufferRef

[pcc]

Post-devmeeting ping.

clang/include/clang/CodeGen/ObjectFilePCHContainerOperations.h

CreatePCHContainerGenerator(CompilerInstance &CI,
                               std::shared_ptr<PCHBuffer> Buffer) const override;
 };
 
 /// A PCHContainerReader implementation that uses LLVM to
 /// wraps Clang modules inside a COFF, ELF, or Mach-O container.
 class ObjectFilePCHContainerReader : public PCHContainerReader {
   StringRef getFormat() const override { return "obj"; }
 
-  /// Initialize an llvm::BitstreamReader with the serialized
-  /// AST inside the PCH container Buffer.
-  void ExtractPCH(llvm::MemoryBufferRef Buffer,
-                  llvm::BitstreamReader &StreamFile) const override;
+  /// Returns the serialized AST inside the PCH container Buffer.
+  StringRef ExtractPCH(llvm::MemoryBufferRef Buffer) const override;
 };
 }
 
 #endif

clang/include/clang/Frontend/PCHContainerOperations.h

 
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringMap.h"
 #include "llvm/Support/MemoryBuffer.h"
 #include <memory>
 
 namespace llvm {
 class raw_pwrite_stream;
-class BitstreamReader;
 }
 
 using llvm::StringRef;
 
 namespace clang {
 
 class ASTConsumer;
 class CodeGenOptions;

 /// containers for serialized ASTs (precompiled headers and clang
 /// modules).
 class PCHContainerReader {
 public: 
   virtual ~PCHContainerReader() = 0;
   /// Equivalent to the format passed to -fmodule-format=
   virtual StringRef getFormat() const = 0;
 
-  /// Initialize an llvm::BitstreamReader with the serialized AST inside
-  /// the PCH container Buffer.
-  virtual void ExtractPCH(llvm::MemoryBufferRef Buffer,
-                          llvm::BitstreamReader &StreamFile) const = 0;
+  /// Returns the serialized AST inside the PCH container Buffer.
+  virtual StringRef ExtractPCH(llvm::MemoryBufferRef Buffer) const = 0;
 };
 
 /// Implements write operations for a raw pass-through PCH container.
 class RawPCHContainerWriter : public PCHContainerWriter {
   StringRef getFormat() const override { return "raw"; }
 
   /// Return an ASTConsumer that can be chained with a
   /// PCHGenerator that writes the module to a flat file.
   std::unique_ptr<ASTConsumer>
   CreatePCHContainerGenerator(CompilerInstance &CI,
                               const std::string &MainFileName,
                               const std::string &OutputFileName,
                               std::unique_ptr<llvm::raw_pwrite_stream> OS,
                               std::shared_ptr<PCHBuffer> Buffer) const override;
 };
 
 /// Implements read operations for a raw pass-through PCH container.
 class RawPCHContainerReader : public PCHContainerReader {
   StringRef getFormat() const override { return "raw"; }
 
-  /// Initialize an llvm::BitstreamReader with Buffer.
-  void ExtractPCH(llvm::MemoryBufferRef Buffer,
-                  llvm::BitstreamReader &StreamFile) const override;
+  /// Simply returns the buffer contained in Buffer.
+  StringRef ExtractPCH(llvm::MemoryBufferRef Buffer) const override;
 };
 
 /// A registry of PCHContainerWriter and -Reader objects for different formats.
 class PCHContainerOperations {
   llvm::StringMap<std::unique_ptr<PCHContainerWriter>> Writers;
   llvm::StringMap<std::unique_ptr<PCHContainerReader>> Readers;
 public:
   /// Automatically registers a RawPCHContainerWriter and

clang/include/clang/Serialization/Module.h

public:
   std::unique_ptr<llvm::MemoryBuffer> Buffer;
 
   /// \brief The size of this file, in bits.
   uint64_t SizeInBits;
 
   /// \brief The global bit offset (or base) of this module
   uint64_t GlobalBitOffset;
 
-  /// \brief The bitstream reader from which we'll read the AST file.
-  llvm::BitstreamReader StreamFile;
+  /// \brief The serialized bitstream data for this file.
+  StringRef Data;
 
   /// \brief The main bitstream cursor for the main block.
   llvm::BitstreamCursor Stream;
 
   /// \brief The source location where the module was explicitly or implicitly
   /// imported in the local translation unit.
   ///
   /// If module A depends on and imports module B, both modules will have the

clang/include/clang/Serialization/ModuleManager.h

enum AddModuleResult {
     /// \brief The module file was just loaded in response to this call.
     NewlyLoaded,
     /// \brief The module file is missing.
     Missing,
     /// \brief The module file is out-of-date.
     OutOfDate
   };
 
-  typedef ASTFileSignature(*ASTFileSignatureReader)(llvm::BitstreamReader &);
+  typedef ASTFileSignature(*ASTFileSignatureReader)(StringRef);
 
   /// \brief Attempts to create a new module and add it to the list of known
   /// modules.
   ///
   /// \param FileName The file name of the module to be loaded.
   ///
   /// \param Type The kind of module being loaded.
   ///

clang/lib/CodeGen/ObjectFilePCHContainerOperations.cpp

ObjectFilePCHContainerWriter::CreatePCHContainerGenerator(
     CompilerInstance &CI, const std::string &MainFileName,
     const std::string &OutputFileName,
     std::unique_ptr<llvm::raw_pwrite_stream> OS,
     std::shared_ptr<PCHBuffer> Buffer) const {
   return llvm::make_unique<PCHContainerGenerator>(
       CI, MainFileName, OutputFileName, std::move(OS), Buffer);
 }
 
-void ObjectFilePCHContainerReader::ExtractPCH(
-    llvm::MemoryBufferRef Buffer, llvm::BitstreamReader &StreamFile) const {
+StringRef
+ObjectFilePCHContainerReader::ExtractPCH(llvm::MemoryBufferRef Buffer) const {
+  StringRef PCH;
   auto OFOrErr = llvm::object::ObjectFile::createObjectFile(Buffer);
   if (OFOrErr) {
     auto &OF = OFOrErr.get();
     bool IsCOFF = isa<llvm::object::COFFObjectFile>(*OF);
     // Find the clang AST section in the container.
     for (auto &Section : OF->sections()) {
       StringRef Name;
       Section.getName(Name);
       if ((!IsCOFF && Name == "__clangast") || (IsCOFF && Name == "clangast")) {
-        StringRef Buf;
-        Section.getContents(Buf);
-        StreamFile = llvm::BitstreamReader(Buf);
-        return;
+        Section.getContents(PCH);
+        return PCH;
       }
     }
   }
   handleAllErrors(OFOrErr.takeError(), [&](const llvm::ErrorInfoBase &EIB) {
     if (EIB.convertToErrorCode() ==
         llvm::object::object_error::invalid_file_type)
       // As a fallback, treat the buffer as a raw AST.
-      StreamFile = llvm::BitstreamReader(Buffer);
+      PCH = Buffer.getBuffer();
     else
       EIB.log(llvm::errs());
   });
+  return PCH;
 }

clang/lib/Frontend/PCHContainerOperations.cpp

 
 std::unique_ptr<ASTConsumer> RawPCHContainerWriter::CreatePCHContainerGenerator(
     CompilerInstance &CI, const std::string &MainFileName,
     const std::string &OutputFileName, std::unique_ptr<llvm::raw_pwrite_stream> OS,
     std::shared_ptr<PCHBuffer> Buffer) const {
   return llvm::make_unique<RawPCHContainerGenerator>(std::move(OS), Buffer);
 }
 
-void RawPCHContainerReader::ExtractPCH(
-    llvm::MemoryBufferRef Buffer, llvm::BitstreamReader &StreamFile) const {
-  StreamFile = llvm::BitstreamReader(Buffer);
+StringRef
+RawPCHContainerReader::ExtractPCH(llvm::MemoryBufferRef Buffer) const {
+  return Buffer.getBuffer();
 }
 
 PCHContainerOperations::PCHContainerOperations() {
   registerWriter(llvm::make_unique<RawPCHContainerWriter>());
   registerReader(llvm::make_unique<RawPCHContainerReader>());
 }

clang/lib/Frontend/SerializedDiagnosticReader.cpp

std::error_code SerializedDiagnosticReader::readDiagnostics(StringRef File) {
   // Open the diagnostics file.
   FileSystemOptions FO;
   FileManager FileMgr(FO);
 
   auto Buffer = FileMgr.getBufferForFile(File);
   if (!Buffer)
     return SDError::CouldNotLoad;
 
-  llvm::BitstreamReader StreamFile(**Buffer);
-  llvm::BitstreamCursor Stream(StreamFile);
+  llvm::BitstreamCursor Stream(**Buffer);
+  Optional<llvm::BitstreamBlockInfo> BlockInfo;
 
   // Sniff for the signature.
   if (Stream.Read(8) != 'D' ||
       Stream.Read(8) != 'I' ||
       Stream.Read(8) != 'A' ||
       Stream.Read(8) != 'G')
     return SDError::InvalidSignature;
 
   // Read the top level blocks.
   while (!Stream.AtEndOfStream()) {
     if (Stream.ReadCode() != llvm::bitc::ENTER_SUBBLOCK)
       return SDError::InvalidDiagnostics;
 
     std::error_code EC;
     switch (Stream.ReadSubBlockID()) {
-    case llvm::bitc::BLOCKINFO_BLOCK_ID:
-      if (Stream.ReadBlockInfoBlock())
+    case llvm::bitc::BLOCKINFO_BLOCK_ID: {
+      BlockInfo = Stream.ReadBlockInfoBlock();
+      if (!BlockInfo)
         return SDError::MalformedBlockInfoBlock;
+      Stream.setBlockInfo(&*BlockInfo);
       continue;
+    }
     case BLOCK_META:
       if ((EC = readMetaBlock(Stream)))
         return EC;
       continue;
     case BLOCK_DIAG:
       if ((EC = readDiagnosticBlock(Stream)))
         return EC;
       continue;

clang/lib/Serialization/ASTReader.cpp

 void ASTReader::ReadDefinedMacros() {
   // Note that we are loading defined macros.
   Deserializing Macros(this);
 
   for (auto &I : llvm::reverse(ModuleMgr)) {
     BitstreamCursor &MacroCursor = I->MacroCursor;
 
     // If there was no preprocessor block, skip this file.
-    if (!MacroCursor.getBitStreamReader())
+    if (MacroCursor.getBitcodeBytes().empty())
       continue;
 
     BitstreamCursor Cursor = MacroCursor;
     Cursor.JumpToBit(I->MacroStartOffset);
 
     RecordData Record;
     while (true) {
       llvm::BitstreamEntry E = Cursor.advanceSkippingSubblocks();
for (unsigned I = 0, N = Loaded.size(); I != N; ++I) {
         updateModuleTimestamp(*M.Mod);
       }
     }
   }
 
   return Success;
 }
 
-static ASTFileSignature readASTFileSignature(llvm::BitstreamReader &StreamFile);
+static ASTFileSignature readASTFileSignature(StringRef PCH);
 
 /// \brief Whether \p Stream starts with the AST/PCH file magic number 'CPCH'.
 static bool startsWithASTFileMagic(BitstreamCursor &Stream) {
   return Stream.canSkipToPos(4) &&
          Stream.Read(8) == 'C' &&
          Stream.Read(8) == 'P' &&
          Stream.Read(8) == 'C' &&
          Stream.Read(8) == 'H';
ASTReader::ReadASTCore(StringRef FileName,
   // module?
   if (FileName != "-") {
     CurrentDir = llvm::sys::path::parent_path(FileName);
     if (CurrentDir.empty()) CurrentDir = ".";
   }
 
   ModuleFile &F = *M;
   BitstreamCursor &Stream = F.Stream;
-  PCHContainerRdr.ExtractPCH(F.Buffer->getMemBufferRef(), F.StreamFile);
-  Stream.init(&F.StreamFile);
+  Stream = BitstreamCursor(PCHContainerRdr.ExtractPCH(*F.Buffer));
   F.SizeInBits = F.Buffer->getBufferSize() * 8;
   
   // Sniff for the signature.
   if (!startsWithASTFileMagic(Stream)) {
     Diag(diag::err_module_file_invalid) << moduleKindForDiagnostic(Type)
                                         << FileName;
     return Failure;
   }
void ASTReader::InitializeContext() {
   }
   ImportedModules.clear();
 }
 
 void ASTReader::finalizeForWriting() {
   // Nothing to do for now.
 }
 
-/// \brief Reads and return the signature record from \p StreamFile's control
-/// block, or else returns 0.
-static ASTFileSignature readASTFileSignature(llvm::BitstreamReader &StreamFile){
-  BitstreamCursor Stream(StreamFile);
+/// \brief Reads and return the signature record from \p PCH's control block, or
+/// else returns 0.
+static ASTFileSignature readASTFileSignature(StringRef PCH) {
+  BitstreamCursor Stream(PCH);
   if (!startsWithASTFileMagic(Stream))
     return 0;
 
   // Scan for the CONTROL_BLOCK_ID block.
   if (SkipCursorToBlock(Stream, CONTROL_BLOCK_ID))
     return 0;
 
   // Scan for SIGNATURE inside the control block.
std::string ASTReader::getOriginalSourceFile(
   auto Buffer = FileMgr.getBufferForFile(ASTFileName);
   if (!Buffer) {
     Diags.Report(diag::err_fe_unable_to_read_pch_file)
         << ASTFileName << Buffer.getError().message();
     return std::string();
   }
 
   // Initialize the stream
-  llvm::BitstreamReader StreamFile;
-  PCHContainerRdr.ExtractPCH((*Buffer)->getMemBufferRef(), StreamFile);
-  BitstreamCursor Stream(StreamFile);
+  BitstreamCursor Stream(PCHContainerRdr.ExtractPCH(**Buffer));
 
   // Sniff for the signature.
   if (!startsWithASTFileMagic(Stream)) {
     Diags.Report(diag::err_fe_not_a_pch_file) << ASTFileName;
     return std::string();
   }
   
   // Scan for the CONTROL_BLOCK_ID block.
bool ASTReader::readASTFileControlBlock(
   // FIXME: This allows use of the VFS; we do not allow use of the
   // VFS when actually loading a module.
   auto Buffer = FileMgr.getBufferForFile(Filename);
   if (!Buffer) {
     return true;
   }
 
   // Initialize the stream
-  llvm::BitstreamReader StreamFile;
-  PCHContainerRdr.ExtractPCH((*Buffer)->getMemBufferRef(), StreamFile);
-  BitstreamCursor Stream(StreamFile);
+  BitstreamCursor Stream(PCHContainerRdr.ExtractPCH(**Buffer));
 
   // Sniff for the signature.
   if (!startsWithASTFileMagic(Stream))
     return true;
 
   // Scan for the CONTROL_BLOCK_ID block.
   if (SkipCursorToBlock(Stream, CONTROL_BLOCK_ID))
     return true;

clang/lib/Serialization/GlobalModuleIndex.cpp

GlobalModuleIndex::readIndex(StringRef Path) {
   llvm::sys::path::append(IndexPath, IndexFileName);
 
   llvm::ErrorOr<std::unique_ptr<llvm::MemoryBuffer>> BufferOrErr =
       llvm::MemoryBuffer::getFile(IndexPath.c_str());
   if (!BufferOrErr)
     return std::make_pair(nullptr, EC_NotFound);
   std::unique_ptr<llvm::MemoryBuffer> Buffer = std::move(BufferOrErr.get());
 
-  /// \brief The bitstream reader from which we'll read the AST file.
-  llvm::BitstreamReader Reader(*Buffer);
-
   /// \brief The main bitstream cursor for the main block.
-  llvm::BitstreamCursor Cursor(Reader);
+  llvm::BitstreamCursor Cursor(*Buffer);
 
   // Sniff for the signature.
   if (Cursor.Read(8) != 'B' ||
       Cursor.Read(8) != 'C' ||
       Cursor.Read(8) != 'G' ||
       Cursor.Read(8) != 'I') {
     return std::make_pair(nullptr, EC_IOError);
   }
bool GlobalModuleIndexBuilder::loadModuleFile(const FileEntry *File) {
   // Open the module file.
 
   auto Buffer = FileMgr.getBufferForFile(File, /*isVolatile=*/true);
   if (!Buffer) {
     return true;
   }
 
   // Initialize the input stream
-  llvm::BitstreamReader InStreamFile;
-  PCHContainerRdr.ExtractPCH((*Buffer)->getMemBufferRef(), InStreamFile);
-  llvm::BitstreamCursor InStream(InStreamFile);
+  llvm::BitstreamCursor InStream(PCHContainerRdr.ExtractPCH(**Buffer));
 
   // Sniff for the signature.
   if (InStream.Read(8) != 'C' ||
       InStream.Read(8) != 'P' ||
       InStream.Read(8) != 'C' ||
       InStream.Read(8) != 'H') {
     return true;
   }

clang/lib/Serialization/ModuleManager.cpp

if (std::unique_ptr<llvm::MemoryBuffer> Buffer = lookupBuffer(FileName)) {
         delete ModuleEntry;
         return Missing;
       }
 
       ModuleEntry->Buffer = std::move(*Buf);
     }
 
     // Initialize the stream.
-    PCHContainerRdr.ExtractPCH(ModuleEntry->Buffer->getMemBufferRef(),
-                               ModuleEntry->StreamFile);
+    ModuleEntry->Data = PCHContainerRdr.ExtractPCH(*ModuleEntry->Buffer);
   }
 
   if (ExpectedSignature) {
     // If we've not read the control block yet, read the signature eagerly now
     // so that we can check it.
     if (!ModuleEntry->Signature)
-      ModuleEntry->Signature = ReadSignature(ModuleEntry->StreamFile);
+      ModuleEntry->Signature = ReadSignature(ModuleEntry->Data);
 
     if (ModuleEntry->Signature != ExpectedSignature) {
       ErrorStr = ModuleEntry->Signature ? "signature mismatch"
                                         : "could not read module signature";
 
       if (NewModule)
         delete ModuleEntry;
       return OutOfDate;

llvm/include/llvm/Bitcode/BitstreamReader.h

 #include <cstdint>
 #include <memory>
 #include <string>
 #include <utility>
 #include <vector>
 
 namespace llvm {
 
-/// This class is used to read from an LLVM bitcode stream, maintaining
-/// information that is global to decoding the entire file. While a file is
-/// being read, multiple cursors can be independently advanced or skipped around
-/// within the file.  These are represented by the BitstreamCursor class.
-class BitstreamReader {
+/// This class maintains the abbreviations read from a block info block.
+class BitstreamBlockInfo {
 public:
   /// This contains information emitted to BLOCKINFO_BLOCK blocks. These
   /// describe abbreviations that all blocks of the specified ID inherit.
   struct BlockInfo {
     unsigned BlockID;
     std::vector<IntrusiveRefCntPtr<BitCodeAbbrev>> Abbrevs;
     std::string Name;
     std::vector<std::pair<unsigned, std::string> > RecordNames;
   };
 
 private:
-  ArrayRef<uint8_t> BitcodeBytes;
-
   std::vector<BlockInfo> BlockInfoRecords;
 
-  /// This is set to true if we don't care about the block/record name
-  /// information in the BlockInfo block. Only llvm-bcanalyzer uses this.
-  bool IgnoreBlockInfoNames = true;
-
 public:
-  BitstreamReader() = default;
-  BitstreamReader(ArrayRef<uint8_t> BitcodeBytes)
-      : BitcodeBytes(BitcodeBytes) {}
-  BitstreamReader(StringRef BitcodeBytes)
-      : BitcodeBytes(reinterpret_cast<const uint8_t *>(BitcodeBytes.data()),
-                     BitcodeBytes.size()) {}
-  BitstreamReader(MemoryBufferRef BitcodeBytes)
-      : BitstreamReader(BitcodeBytes.getBuffer()) {}
-
-  ArrayRef<uint8_t> getBitcodeBytes() { return BitcodeBytes; }
-
-  /// This is called by clients that want block/record name information.
-  void CollectBlockInfoNames() { IgnoreBlockInfoNames = false; }
-  bool isIgnoringBlockInfoNames() { return IgnoreBlockInfoNames; }
-
-  //===--------------------------------------------------------------------===//
-  // Block Manipulation
-  //===--------------------------------------------------------------------===//
-
-  /// Return true if we've already read and processed the block info block for
-  /// this Bitstream. We only process it for the first cursor that walks over
-  /// it.
-  bool hasBlockInfoRecords() const { return !BlockInfoRecords.empty(); }
-
   /// If there is block info for the specified ID, return it, otherwise return
   /// null.
   const BlockInfo *getBlockInfo(unsigned BlockID) const {
     // Common case, the most recent entry matches BlockID.
     if (!BlockInfoRecords.empty() && BlockInfoRecords.back().BlockID == BlockID)
       return &BlockInfoRecords.back();
 
     for (unsigned i = 0, e = static_cast<unsigned>(BlockInfoRecords.size());
          i != e; ++i)
       if (BlockInfoRecords[i].BlockID == BlockID)
         return &BlockInfoRecords[i];
     return nullptr;
   }
 
   BlockInfo &getOrCreateBlockInfo(unsigned BlockID) {
     if (const BlockInfo *BI = getBlockInfo(BlockID))
       return *const_cast<BlockInfo*>(BI);
 
     // Otherwise, add a new record.
     BlockInfoRecords.emplace_back();
     BlockInfoRecords.back().BlockID = BlockID;
     return BlockInfoRecords.back();
   }
-
-  /// Takes block info from the other bitstream reader.
-  ///
-  /// This is a "take" operation because BlockInfo records are non-trivial, and
-  /// indeed rather expensive.
-  void takeBlockInfo(BitstreamReader &&Other) {
-    assert(!hasBlockInfoRecords());
-    BlockInfoRecords = std::move(Other.BlockInfoRecords);
-  }
 };
 
 /// This represents a position within a bitstream. There may be multiple
 /// independent cursors reading within one bitstream, each maintaining their
 /// own local state.
 class SimpleBitstreamCursor {
-  BitstreamReader *R = nullptr;
+  ArrayRef<uint8_t> BitcodeBytes;
   size_t NextChar = 0;
 
 public:
   /// This is the current data we have pulled from the stream but have not
   /// returned to the client. This is specifically and intentionally defined to
   /// follow the word size of the host machine for efficiency. We use word_t in
   /// places that are aware of this to make it perfectly explicit what is going
   /// on.
   typedef size_t word_t;
 
 private:
   word_t CurWord = 0;
 
   /// This is the number of bits in CurWord that are valid. This is always from
   /// [0...bits_of(size_t)-1] inclusive.
   unsigned BitsInCurWord = 0;
 
 public:
   static const size_t MaxChunkSize = sizeof(word_t) * 8;
 
   SimpleBitstreamCursor() = default;
-
-  explicit SimpleBitstreamCursor(BitstreamReader &R) : R(&R) {}
-  explicit SimpleBitstreamCursor(BitstreamReader *R) : R(R) {}
+  explicit SimpleBitstreamCursor(ArrayRef<uint8_t> BitcodeBytes)
+      : BitcodeBytes(BitcodeBytes) {}
+  explicit SimpleBitstreamCursor(StringRef BitcodeBytes)
+      : BitcodeBytes(reinterpret_cast<const uint8_t *>(BitcodeBytes.data()),
+                     BitcodeBytes.size()) {}
+  explicit SimpleBitstreamCursor(MemoryBufferRef BitcodeBytes)
+      : SimpleBitstreamCursor(BitcodeBytes.getBuffer()) {}
 
   bool canSkipToPos(size_t pos) const {
     // pos can be skipped to if it is a valid address or one byte past the end.
-    return pos <= R->getBitcodeBytes().size();
+    return pos <= BitcodeBytes.size();
   }
 
   bool AtEndOfStream() {
-    return BitsInCurWord == 0 && R->getBitcodeBytes().size() <= NextChar;
+    return BitsInCurWord == 0 && BitcodeBytes.size() <= NextChar;
   }
 
   /// Return the bit # of the bit we are reading.
   uint64_t GetCurrentBitNo() const {
     return NextChar*CHAR_BIT - BitsInCurWord;
   }
 
   // Return the byte # of the current bit.
   uint64_t getCurrentByteNo() const { return GetCurrentBitNo() / 8; }
 
-  BitstreamReader *getBitStreamReader() { return R; }
-  const BitstreamReader *getBitStreamReader() const { return R; }
+  ArrayRef<uint8_t> getBitcodeBytes() const { return BitcodeBytes; }
 
   /// Reset the stream to the specified bit number.
   void JumpToBit(uint64_t BitNo) {
     size_t ByteNo = size_t(BitNo/8) & ~(sizeof(word_t)-1);
     unsigned WordBitNo = unsigned(BitNo & (sizeof(word_t)*8-1));
     assert(canSkipToPos(ByteNo) && "Invalid location");
 
     // Move the cursor to the right word.
     NextChar = ByteNo;
     BitsInCurWord = 0;
 
     // Skip over any bits that are already consumed.
     if (WordBitNo)
       Read(WordBitNo);
   }
 
-  /// Reset the stream to the bit pointed at by the specified pointer.
-  ///
-  /// The pointer must be a dereferenceable pointer into the bytes in the
-  /// underlying memory object.
-  void jumpToPointer(const uint8_t *Pointer) {
-    auto *Pointer0 = getPointerToByte(0, 1);
-    assert((intptr_t)Pointer0 <= (intptr_t)Pointer &&
-           "Expected pointer into bitstream");
-
-    JumpToBit(8 * (Pointer - Pointer0));
-    assert((intptr_t)getPointerToByte(getCurrentByteNo(), 1) ==
-               (intptr_t)Pointer &&
-           "Expected to reach pointer");
-  }
-  void jumpToPointer(const char *Pointer) {
-    jumpToPointer((const uint8_t *)Pointer);
-  }
-
   /// Get a pointer into the bitstream at the specified byte offset.
   const uint8_t *getPointerToByte(uint64_t ByteNo, uint64_t NumBytes) {
-    return R->getBitcodeBytes().data() + ByteNo;
+    return BitcodeBytes.data() + ByteNo;
   }
 
   /// Get a pointer into the bitstream at the specified bit offset.
   ///
   /// The bit offset must be on a byte boundary.
   const uint8_t *getPointerToBit(uint64_t BitNo, uint64_t NumBytes) {
     assert(!(BitNo % 8) && "Expected bit on byte boundary");
     return getPointerToByte(BitNo / 8, NumBytes);
   }
 
   void fillCurWord() {
-    ArrayRef<uint8_t> Buf = R->getBitcodeBytes();
-    if (NextChar >= Buf.size())
+    if (NextChar >= BitcodeBytes.size())
       report_fatal_error("Unexpected end of file");
 
     // Read the next word from the stream.
-    const uint8_t *NextCharPtr = Buf.data() + NextChar;
+    const uint8_t *NextCharPtr = BitcodeBytes.data() + NextChar;
     unsigned BytesRead;
-    if (Buf.size() >= NextChar + sizeof(word_t)) {
+    if (BitcodeBytes.size() >= NextChar + sizeof(word_t)) {
       BytesRead = sizeof(word_t);
       CurWord =
           support::endian::read<word_t, support::little, support::unaligned>(
               NextCharPtr);
     } else {
       // Short read.
-      BytesRead = Buf.size() - NextChar;
+      BytesRead = BitcodeBytes.size() - NextChar;
       CurWord = 0;
       for (unsigned B = 0; B != BytesRead; ++B)
         CurWord |= uint64_t(NextCharPtr[B]) << (B * 8);
     }
     NextChar += BytesRead;
     BitsInCurWord = BytesRead * 8;
   }
 
if (sizeof(word_t) > 4 &&
       BitsInCurWord = 32;
       return;
     }
 
     BitsInCurWord = 0;
   }
 
   /// Skip to the end of the file.
-  void skipToEnd() { NextChar = R->getBitcodeBytes().size(); }
+  void skipToEnd() { NextChar = BitcodeBytes.size(); }
 };
 
 /// When advancing through a bitstream cursor, each advance can discover a few
 /// different kinds of entries:
 struct BitstreamEntry {
   enum {
     Error,    // Malformed bitcode was found.
     EndBlock, // We've reached the end of the current block, (or the end of the
struct Block {
     std::vector<IntrusiveRefCntPtr<BitCodeAbbrev>> PrevAbbrevs;
 
     explicit Block(unsigned PCS) : PrevCodeSize(PCS) {}
   };
 
   /// This tracks the codesize of parent blocks.
   SmallVector<Block, 8> BlockScope;
 
+  BitstreamBlockInfo *BlockInfo = nullptr;
+
 public:
   static const size_t MaxChunkSize = sizeof(word_t) * 8;
 
   BitstreamCursor() = default;
-
-  explicit BitstreamCursor(BitstreamReader &R) { init(&R); }
-
-  void init(BitstreamReader *R) {
-    freeState();
-    SimpleBitstreamCursor::operator=(SimpleBitstreamCursor(R));
-    CurCodeSize = 2;
-  }
-
-  void freeState();
+  explicit BitstreamCursor(ArrayRef<uint8_t> BitcodeBytes)
+      : SimpleBitstreamCursor(BitcodeBytes) {}
+  explicit BitstreamCursor(StringRef BitcodeBytes)
+      : SimpleBitstreamCursor(BitcodeBytes) {}
+  explicit BitstreamCursor(MemoryBufferRef BitcodeBytes)
+      : SimpleBitstreamCursor(BitcodeBytes) {}
 
   using SimpleBitstreamCursor::canSkipToPos;
   using SimpleBitstreamCursor::AtEndOfStream;
+  using SimpleBitstreamCursor::getBitcodeBytes;
   using SimpleBitstreamCursor::GetCurrentBitNo;
   using SimpleBitstreamCursor::getCurrentByteNo;
   using SimpleBitstreamCursor::getPointerToByte;
-  using SimpleBitstreamCursor::getBitStreamReader;
   using SimpleBitstreamCursor::JumpToBit;
   using SimpleBitstreamCursor::fillCurWord;
   using SimpleBitstreamCursor::Read;
   using SimpleBitstreamCursor::ReadVBR;
   using SimpleBitstreamCursor::ReadVBR64;
 
   /// Return the number of bits used to encode an abbrev #.
   unsigned getAbbrevIDWidth() const { return CurCodeSize; }
public:
   unsigned readRecord(unsigned AbbrevID, SmallVectorImpl<uint64_t> &Vals,
                       StringRef *Blob = nullptr);
 
   //===--------------------------------------------------------------------===//
   // Abbrev Processing
   //===--------------------------------------------------------------------===//
   void ReadAbbrevRecord();
 
-  bool ReadBlockInfoBlock();
+  /// Read and return a block info block from the bitstream. If an error was
+  /// encountered, return None.
+  ///
+  /// \param ReadBlockInfoNames Whether to read block/record name information in
+  /// the BlockInfo block. Only llvm-bcanalyzer uses this.
+  Optional<BitstreamBlockInfo>
+  ReadBlockInfoBlock(bool ReadBlockInfoNames = false);
+
+  /// Set the block info to be used by this BitstreamCursor to interpret
+  /// abbreviated records.
+  void setBlockInfo(BitstreamBlockInfo *BI) { BlockInfo = BI; }
 };
 
 } // end llvm namespace
 
 #endif // LLVM_BITCODE_BITSTREAMREADER_H

llvm/lib/Bitcode/Reader/BitcodeReader.cpp

private:
   Metadata *resolveTypeRefArray(Metadata *MaybeTuple);
 };
 
 class BitcodeReaderBase {
 protected:
   BitcodeReaderBase(MemoryBuffer *Buffer) : Buffer(Buffer) {}
 
   std::unique_ptr<MemoryBuffer> Buffer;
-  std::unique_ptr<BitstreamReader> StreamFile;
+  BitstreamBlockInfo BlockInfo;
   BitstreamCursor Stream;
 
   std::error_code initStream();
+  bool readBlockInfo();
 
   virtual std::error_code error(const Twine &Message) = 0;
   virtual ~BitcodeReaderBase() = default;
 };
 
 std::error_code BitcodeReaderBase::initStream() {
   const unsigned char *BufPtr = (const unsigned char*)Buffer->getBufferStart();
   const unsigned char *BufEnd = BufPtr+Buffer->getBufferSize();
 
   if (Buffer->getBufferSize() & 3)
     return error("Invalid bitcode signature");
 
   // If we have a wrapper header, parse it and ignore the non-bc file contents.
   // The magic number is 0x0B17C0DE stored in little endian.
   if (isBitcodeWrapper(BufPtr, BufEnd))
     if (SkipBitcodeWrapperHeader(BufPtr, BufEnd, true))
       return error("Invalid bitcode wrapper header");
 
-  StreamFile.reset(new BitstreamReader(ArrayRef<uint8_t>(BufPtr, BufEnd)));
-  Stream.init(&*StreamFile);
+  Stream = BitstreamCursor(ArrayRef<uint8_t>(BufPtr, BufEnd));
+  Stream.setBlockInfo(&BlockInfo);
 
   return std::error_code();
 }
 
 class BitcodeReader : public BitcodeReaderBase, public GVMaterializer {
   LLVMContext &Context;
   Module *TheModule = nullptr;
   // Next offset to start scanning for lazy parsing of function bodies.
std::error_code BitcodeReader::parseMetadataStrings(ArrayRef<uint64_t> Record,
   unsigned NumStrings = Record[0];
   unsigned StringsOffset = Record[1];
   if (!NumStrings)
     return error("Invalid record: metadata strings with no strings");
   if (StringsOffset > Blob.size())
     return error("Invalid record: metadata strings corrupt offset");
 
   StringRef Lengths = Blob.slice(0, StringsOffset);
-  SimpleBitstreamCursor R(*StreamFile);
-  R.jumpToPointer(Lengths.begin());
+  SimpleBitstreamCursor R(Lengths);
 
   StringRef Strings = Blob.drop_front(StringsOffset);
   do {
     if (R.AtEndOfStream())
       return error("Invalid record: metadata strings bad length");
 
     unsigned Size = R.ReadVBR(6);
     if (Strings.size() < Size)
case bitc::IDENTIFICATION_CODE_EPOCH: { // EPOCH:      [epoch#]
           Twine("Incompatible epoch: Bitcode '") + Twine(epoch) +
           "' vs current: '" + Twine(bitc::BITCODE_CURRENT_EPOCH) + "'");
       }
     }
     }
   }
 }
 
+bool BitcodeReaderBase::readBlockInfo() {
+  Optional<BitstreamBlockInfo> NewBlockInfo = Stream.ReadBlockInfoBlock();
+  if (!NewBlockInfo)
+    return true;
+  BlockInfo = std::move(*NewBlockInfo);
+  return false;
+}
+
 std::error_code BitcodeReader::parseModule(uint64_t ResumeBit,
                                            bool ShouldLazyLoadMetadata) {
   if (ResumeBit)
     Stream.JumpToBit(ResumeBit);
   else if (Stream.EnterSubBlock(bitc::MODULE_BLOCK_ID))
     return error("Invalid record");
 
   SmallVector<uint64_t, 64> Record;
while (true) {
 
     case BitstreamEntry::SubBlock:
       switch (Entry.ID) {
       default:  // Skip unknown content.
         if (Stream.SkipBlock())
           return error("Invalid record");
         break;
       case bitc::BLOCKINFO_BLOCK_ID:
-        if (Stream.ReadBlockInfoBlock())
+        if (readBlockInfo())
           return error("Malformed block");
         break;
       case bitc::PARAMATTR_BLOCK_ID:
         if (std::error_code EC = parseAttributeBlock())
           return EC;
         break;
       case bitc::PARAMATTR_GROUP_BLOCK_ID:
         if (std::error_code EC = parseAttributeGroupBlock())
case BitstreamEntry::SubBlock:
       }
       switch (Entry.ID) {
       default: // Skip unknown content.
         if (Stream.SkipBlock())
           return error("Invalid record");
         break;
       case bitc::BLOCKINFO_BLOCK_ID:
         // Need to parse these to get abbrev ids (e.g. for VST)
-        if (Stream.ReadBlockInfoBlock())
+        if (readBlockInfo())
           return error("Malformed block");
         break;
       case bitc::VALUE_SYMTAB_BLOCK_ID:
         // Should have been parsed earlier via VSTOffset, unless there
         // is no summary section.
         assert(((SeenValueSymbolTable && VSTOffset > 0) ||
                 !SeenGlobalValSummary) &&
                "Expected early VST parse via VSTOffset record");

llvm/lib/Bitcode/Reader/BitstreamReader.cpp

 #include <string>
 
 using namespace llvm;
 
 //===----------------------------------------------------------------------===//
 //  BitstreamCursor implementation
 //===----------------------------------------------------------------------===//
 
-void BitstreamCursor::freeState() {
-  // Free all the Abbrevs.
-  CurAbbrevs.clear();
-
-  // Free all the Abbrevs in the block scope.
-  BlockScope.clear();
-}
-
 /// EnterSubBlock - Having read the ENTER_SUBBLOCK abbrevid, enter
 /// the block, and return true if the block has an error.
 bool BitstreamCursor::EnterSubBlock(unsigned BlockID, unsigned *NumWordsP) {
   // Save the current block's state on BlockScope.
   BlockScope.push_back(Block(CurCodeSize));
   BlockScope.back().PrevAbbrevs.swap(CurAbbrevs);
 
   // Add the abbrevs specific to this block to the CurAbbrevs list.
-  if (const BitstreamReader::BlockInfo *Info =
-          getBitStreamReader()->getBlockInfo(BlockID)) {
+  if (BlockInfo) {
+    if (const BitstreamBlockInfo::BlockInfo *Info =
+            BlockInfo->getBlockInfo(BlockID)) {
       CurAbbrevs.insert(CurAbbrevs.end(), Info->Abbrevs.begin(),
                         Info->Abbrevs.end());
     }
+  }
 
   // Get the codesize of this block.
   CurCodeSize = ReadVBR(bitc::CodeLenWidth);
   // We can't read more than MaxChunkSize at a time
   if (CurCodeSize > MaxChunkSize)
     return true;
 
   SkipToFourByteBoundary();
if (BitCodeAbbrevOp::hasEncodingData(E)) {
       Abbv->Add(BitCodeAbbrevOp(E));
   }
 
   if (Abbv->getNumOperandInfos() == 0)
     report_fatal_error("Abbrev record with no operands");
   CurAbbrevs.push_back(Abbv);
 }
 
-bool BitstreamCursor::ReadBlockInfoBlock() {
-  // We expect the client to read the block info block at most once.
-  if (getBitStreamReader()->hasBlockInfoRecords())
-    report_fatal_error("Duplicate read of block info block");
+Optional<BitstreamBlockInfo>
+BitstreamCursor::ReadBlockInfoBlock(bool ReadBlockInfoNames) {
+  if (EnterSubBlock(bitc::BLOCKINFO_BLOCK_ID)) return None;
 
-  if (EnterSubBlock(bitc::BLOCKINFO_BLOCK_ID)) return true;
+  BitstreamBlockInfo NewBlockInfo;
 
   SmallVector<uint64_t, 64> Record;
-  BitstreamReader::BlockInfo *CurBlockInfo = nullptr;
+  BitstreamBlockInfo::BlockInfo *CurBlockInfo = nullptr;
 
   // Read all the records for this module.
   while (true) {
     BitstreamEntry Entry = advanceSkippingSubblocks(AF_DontAutoprocessAbbrevs);
 
     switch (Entry.Kind) {
     case llvm::BitstreamEntry::SubBlock: // Handled for us already.
     case llvm::BitstreamEntry::Error:
-      return true;
+      return None;
     case llvm::BitstreamEntry::EndBlock:
-      return false;
+      return std::move(NewBlockInfo);
     case llvm::BitstreamEntry::Record:
       // The interesting case.
       break;
     }
 
     // Read abbrev records, associate them with CurBID.
     if (Entry.ID == bitc::DEFINE_ABBREV) {
-      if (!CurBlockInfo) return true;
+      if (!CurBlockInfo) return None;
       ReadAbbrevRecord();
 
       // ReadAbbrevRecord installs the abbrev in CurAbbrevs.  Move it to the
       // appropriate BlockInfo.
       CurBlockInfo->Abbrevs.push_back(std::move(CurAbbrevs.back()));
       CurAbbrevs.pop_back();
       continue;
     }
 
     // Read a record.
     Record.clear();
     switch (readRecord(Entry.ID, Record)) {
       default: break;  // Default behavior, ignore unknown content.
       case bitc::BLOCKINFO_CODE_SETBID:
-        if (Record.size() < 1) return true;
-        CurBlockInfo =
-            &getBitStreamReader()->getOrCreateBlockInfo((unsigned)Record[0]);
+        if (Record.size() < 1) return None;
+        CurBlockInfo = &NewBlockInfo.getOrCreateBlockInfo((unsigned)Record[0]);
         break;
       case bitc::BLOCKINFO_CODE_BLOCKNAME: {
-        if (!CurBlockInfo) return true;
-        if (getBitStreamReader()->isIgnoringBlockInfoNames())
+        if (!CurBlockInfo) return None;
+        if (!ReadBlockInfoNames)
           break; // Ignore name.
         std::string Name;
         for (unsigned i = 0, e = Record.size(); i != e; ++i)
           Name += (char)Record[i];
         CurBlockInfo->Name = Name;
         break;
       }
       case bitc::BLOCKINFO_CODE_SETRECORDNAME: {
-        if (!CurBlockInfo) return true;
-        if (getBitStreamReader()->isIgnoringBlockInfoNames())
+        if (!CurBlockInfo) return None;
+        if (!ReadBlockInfoNames)
           break; // Ignore name.
         std::string Name;
         for (unsigned i = 1, e = Record.size(); i != e; ++i)
           Name += (char)Record[i];
         CurBlockInfo->RecordNames.push_back(std::make_pair((unsigned)Record[0],
                                                            Name));
         break;
       }
     }
   }
 }

llvm/tools/llvm-bcanalyzer/llvm-bcanalyzer.cpp

enum CurStreamTypeType {
   LLVMIRBitstream
 };
 
 }
 
 /// GetBlockName - Return a symbolic block name if known, otherwise return
 /// null.
 static const char *GetBlockName(unsigned BlockID,
-                                const BitstreamReader &StreamFile,
+                                const BitstreamBlockInfo &BlockInfo,
                                 CurStreamTypeType CurStreamType) {
   // Standard blocks for all bitcode files.
   if (BlockID < bitc::FIRST_APPLICATION_BLOCKID) {
     if (BlockID == bitc::BLOCKINFO_BLOCK_ID)
       return "BLOCKINFO_BLOCK";
     return nullptr;
   }
 
   // Check to see if we have a blockinfo record for this block, with a name.
-  if (const BitstreamReader::BlockInfo *Info =
-        StreamFile.getBlockInfo(BlockID)) {
+  if (const BitstreamBlockInfo::BlockInfo *Info =
+          BlockInfo.getBlockInfo(BlockID)) {
     if (!Info->Name.empty())
       return Info->Name.c_str();
   }
 
 
   if (CurStreamType != LLVMIRBitstream) return nullptr;
 
   switch (BlockID) {
case bitc::GLOBALVAL_SUMMARY_BLOCK_ID:
                                            return "GLOBALVAL_SUMMARY_BLOCK";
   case bitc::MODULE_STRTAB_BLOCK_ID:       return "MODULE_STRTAB_BLOCK";
   }
 }
 
 /// GetCodeName - Return a symbolic code name if known, otherwise return
 /// null.
 static const char *GetCodeName(unsigned CodeID, unsigned BlockID,
-                               const BitstreamReader &StreamFile,
+                               const BitstreamBlockInfo &BlockInfo,
                                CurStreamTypeType CurStreamType) {
   // Standard blocks for all bitcode files.
   if (BlockID < bitc::FIRST_APPLICATION_BLOCKID) {
     if (BlockID == bitc::BLOCKINFO_BLOCK_ID) {
       switch (CodeID) {
       default: return nullptr;
       case bitc::BLOCKINFO_CODE_SETBID:        return "SETBID";
       case bitc::BLOCKINFO_CODE_BLOCKNAME:     return "BLOCKNAME";
       case bitc::BLOCKINFO_CODE_SETRECORDNAME: return "SETRECORDNAME";
       }
     }
     return nullptr;
   }
 
   // Check to see if we have a blockinfo record for this record, with a name.
-  if (const BitstreamReader::BlockInfo *Info =
-        StreamFile.getBlockInfo(BlockID)) {
+  if (const BitstreamBlockInfo::BlockInfo *Info =
+        BlockInfo.getBlockInfo(BlockID)) {
     for (unsigned i = 0, e = Info->RecordNames.size(); i != e; ++i)
       if (Info->RecordNames[i].first == CodeID)
         return Info->RecordNames[i].second.c_str();
   }
 
 
   if (CurStreamType != LLVMIRBitstream) return nullptr;
 

 
 /// ReportError - All bitcode analysis errors go through this function, making this a
 /// good place to breakpoint if debugging.
 static bool ReportError(const Twine &Err) {
   errs() << Err << "\n";
   return true;
 }
 
-static bool decodeMetadataStringsBlob(BitstreamReader &Reader, StringRef Indent,
+static bool decodeMetadataStringsBlob(StringRef Indent,
                                       ArrayRef<uint64_t> Record,
                                       StringRef Blob) {
   if (Blob.empty())
     return true;
 
   if (Record.size() != 2)
     return true;
 
   unsigned NumStrings = Record[0];
   unsigned StringsOffset = Record[1];
   outs() << " num-strings = " << NumStrings << " {\n";
 
   StringRef Lengths = Blob.slice(0, StringsOffset);
-  SimpleBitstreamCursor R(Reader);
-  R.jumpToPointer(Lengths.begin());
-
+  SimpleBitstreamCursor R(Lengths);
   StringRef Strings = Blob.drop_front(StringsOffset);
   do {
     if (R.AtEndOfStream())
       return ReportError("bad length");
 
     unsigned Size = R.ReadVBR(6);
     if (Strings.size() < Size)
       return ReportError("truncated chars");
 
     outs() << Indent << "    '";
     outs().write_escaped(Strings.slice(0, Size), /*hex=*/true);
     outs() << "'\n";
     Strings = Strings.drop_front(Size);
   } while (--NumStrings);
 
   outs() << Indent << "  }";
   return false;
 }
 
-static bool decodeBlob(unsigned Code, unsigned BlockID, BitstreamReader &Reader,
-                       StringRef Indent, ArrayRef<uint64_t> Record,
-                       StringRef Blob) {
+static bool decodeBlob(unsigned Code, unsigned BlockID, StringRef Indent,
+                       ArrayRef<uint64_t> Record, StringRef Blob) {
   if (BlockID != bitc::METADATA_BLOCK_ID)
     return true;
   if (Code != bitc::METADATA_STRINGS)
     return true;
 
-  return decodeMetadataStringsBlob(Reader, Indent, Record, Blob);
+  return decodeMetadataStringsBlob(Indent, Record, Blob);
 }
 
 /// ParseBlock - Read a block, updating statistics, etc.
-static bool ParseBlock(BitstreamCursor &Stream, unsigned BlockID,
-                       unsigned IndentLevel, CurStreamTypeType CurStreamType) {
+static bool ParseBlock(BitstreamCursor &Stream, BitstreamBlockInfo &BlockInfo,
+                       unsigned BlockID, unsigned IndentLevel,
+                       CurStreamTypeType CurStreamType) {
   std::string Indent(IndentLevel*2, ' ');
   uint64_t BlockBitStart = Stream.GetCurrentBitNo();
 
   // Get the statistics for this BlockID.
   PerBlockIDStats &BlockStats = BlockIDStats[BlockID];
 
   BlockStats.NumInstances++;
 
   // BLOCKINFO is a special part of the stream.
   bool DumpRecords = Dump;
   if (BlockID == bitc::BLOCKINFO_BLOCK_ID) {
     if (Dump) outs() << Indent << "<BLOCKINFO_BLOCK/>\n";
-    if (BitstreamCursor(Stream).ReadBlockInfoBlock())
+    Optional<BitstreamBlockInfo> NewBlockInfo =
+        Stream.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true);
+    if (!NewBlockInfo)
       return ReportError("Malformed BlockInfoBlock");
+    BlockInfo = std::move(*NewBlockInfo);
+    Stream.JumpToBit(BlockBitStart);
     // It's not really interesting to dump the contents of the blockinfo block.
     DumpRecords = false;
   }
 
   unsigned NumWords = 0;
   if (Stream.EnterSubBlock(BlockID, &NumWords))
     return ReportError("Malformed block record");
 
   // Keep it for later, when we see a MODULE_HASH record
   uint64_t BlockEntryPos = Stream.getCurrentByteNo();
 
   const char *BlockName = nullptr;
   if (DumpRecords) {
     outs() << Indent << "<";
-    if ((BlockName = GetBlockName(BlockID, *Stream.getBitStreamReader(),
-                                  CurStreamType)))
+    if ((BlockName = GetBlockName(BlockID, BlockInfo, CurStreamType)))
       outs() << BlockName;
     else
       outs() << "UnknownBlock" << BlockID;
 
     if (NonSymbolic && BlockName)
       outs() << " BlockID=" << BlockID;
 
     outs() << " NumWords=" << NumWords
case BitstreamEntry::EndBlock: {
         else
           outs() << "UnknownBlock" << BlockID << ">\n";
       }
       return false;
     }
         
     case BitstreamEntry::SubBlock: {
       uint64_t SubBlockBitStart = Stream.GetCurrentBitNo();
-      if (ParseBlock(Stream, Entry.ID, IndentLevel+1, CurStreamType))
+      if (ParseBlock(Stream, BlockInfo, Entry.ID, IndentLevel + 1,
+                     CurStreamType))
         return true;
       ++BlockStats.NumSubBlocks;
       uint64_t SubBlockBitEnd = Stream.GetCurrentBitNo();
       
       // Don't include subblock sizes in the size of this block.
       BlockBitStart += SubBlockBitEnd-SubBlockBitStart;
       continue;
     }
while (1) {
     if (Entry.ID != bitc::UNABBREV_RECORD) {
       BlockStats.CodeFreq[Code].NumAbbrev++;
       ++BlockStats.NumAbbreviatedRecords;
     }
 
     if (DumpRecords) {
       outs() << Indent << "  <";
       if (const char *CodeName =
-            GetCodeName(Code, BlockID, *Stream.getBitStreamReader(),
-                        CurStreamType))
+              GetCodeName(Code, BlockID, BlockInfo, CurStreamType))
         outs() << CodeName;
       else
         outs() << "UnknownCode" << Code;
-      if (NonSymbolic &&
-          GetCodeName(Code, BlockID, *Stream.getBitStreamReader(),
-                      CurStreamType))
+      if (NonSymbolic && GetCodeName(Code, BlockID, BlockInfo, CurStreamType))
         outs() << " codeid=" << Code;
       const BitCodeAbbrev *Abbv = nullptr;
       if (Entry.ID != bitc::UNABBREV_RECORD) {
         Abbv = Stream.getAbbrev(Entry.ID);
         outs() << " abbrevid=" << Entry.ID;
       }
 
       for (unsigned i = 0, e = Record.size(); i != e; ++i)
if (DumpRecords) {
             Str += (char)Record[j];
           }
           if (ArrayIsPrintable)
             outs() << " record string = '" << Str << "'";
           break;
         }
       }
 
-      if (Blob.data() && decodeBlob(Code, BlockID, *Stream.getBitStreamReader(),
-                                    Indent, Record, Blob)) {
+      if (Blob.data() && decodeBlob(Code, BlockID, Indent, Record, Blob)) {
         outs() << " blob data = ";
         if (ShowBinaryBlobs) {
           outs() << "'";
           outs().write_escaped(Blob, /*hex=*/true) << "'";
         } else {
           bool BlobIsPrintable = true;
           for (unsigned i = 0, e = Blob.size(); i != e; ++i)
             if (!isprint(static_cast<unsigned char>(Blob[i]))) {

 }
 static void PrintSize(uint64_t Bits) {
   outs() << format("%lub/%.2fB/%luW", (unsigned long)Bits,
                    (double)Bits/8, (unsigned long)(Bits/32));
 }
 
 static bool openBitcodeFile(StringRef Path,
                             std::unique_ptr<MemoryBuffer> &MemBuf,
-                            BitstreamReader &StreamFile,
                             BitstreamCursor &Stream,
                             CurStreamTypeType &CurStreamType) {
   // Read the input file.
   ErrorOr<std::unique_ptr<MemoryBuffer>> MemBufOrErr =
       MemoryBuffer::getFileOrSTDIN(Path);
   if (std::error_code EC = MemBufOrErr.getError())
     return ReportError(Twine("ReportError reading '") + Path + "': " + EC.message());
   MemBuf = std::move(MemBufOrErr.get());
if (Dump) {
              << " Size=" << format_hex(Size, 10)
              << " CPUType=" << format_hex(CPUType, 10) << "/>\n";
     }
 
     if (SkipBitcodeWrapperHeader(BufPtr, EndBufPtr, true))
       return ReportError("Invalid bitcode wrapper header");
   }
 
-  StreamFile = BitstreamReader(ArrayRef<uint8_t>(BufPtr, EndBufPtr));
-  Stream = BitstreamCursor(StreamFile);
-  StreamFile.CollectBlockInfoNames();
+  Stream = BitstreamCursor(ArrayRef<uint8_t>(BufPtr, EndBufPtr));
 
   // Read the stream signature.
   char Signature[6];
   Signature[0] = Stream.Read(8);
   Signature[1] = Stream.Read(8);
   Signature[2] = Stream.Read(4);
   Signature[3] = Stream.Read(4);
   Signature[4] = Stream.Read(4);
   Signature[5] = Stream.Read(4);
 
   // Autodetect the file contents, if it is one we know.
   CurStreamType = UnknownBitstream;
   if (Signature[0] == 'B' && Signature[1] == 'C' &&
       Signature[2] == 0x0 && Signature[3] == 0xC &&
       Signature[4] == 0xE && Signature[5] == 0xD)
     CurStreamType = LLVMIRBitstream;
 
   return false;
 }
 
 /// AnalyzeBitcode - Analyze the bitcode file specified by InputFilename.
 static int AnalyzeBitcode() {
   std::unique_ptr<MemoryBuffer> StreamBuffer;
-  BitstreamReader StreamFile;
   BitstreamCursor Stream;
+  BitstreamBlockInfo BlockInfo;
   CurStreamTypeType CurStreamType;
-  if (openBitcodeFile(InputFilename, StreamBuffer, StreamFile, Stream,
-                      CurStreamType))
+  if (openBitcodeFile(InputFilename, StreamBuffer, Stream, CurStreamType))
     return true;
+  Stream.setBlockInfo(&BlockInfo);
 
   // Read block info from BlockInfoFilename, if specified.
   // The block info must be a top-level block.
   if (!BlockInfoFilename.empty()) {
     std::unique_ptr<MemoryBuffer> BlockInfoBuffer;
-    BitstreamReader BlockInfoFile;
     BitstreamCursor BlockInfoCursor;
     CurStreamTypeType BlockInfoStreamType;
-    if (openBitcodeFile(BlockInfoFilename, BlockInfoBuffer, BlockInfoFile,
-                        BlockInfoCursor, BlockInfoStreamType))
+    if (openBitcodeFile(BlockInfoFilename, BlockInfoBuffer, BlockInfoCursor,
+                        BlockInfoStreamType))
       return true;
 
     while (!BlockInfoCursor.AtEndOfStream()) {
       unsigned Code = BlockInfoCursor.ReadCode();
       if (Code != bitc::ENTER_SUBBLOCK)
         return ReportError("Invalid record at top-level in block info file");
 
       unsigned BlockID = BlockInfoCursor.ReadSubBlockID();
       if (BlockID == bitc::BLOCKINFO_BLOCK_ID) {
-        if (BlockInfoCursor.ReadBlockInfoBlock())
+        Optional<BitstreamBlockInfo> NewBlockInfo =
+            BlockInfoCursor.ReadBlockInfoBlock(/*ReadBlockInfoNames=*/true);
+        if (!NewBlockInfo)
           return ReportError("Malformed BlockInfoBlock in block info file");
+        BlockInfo = std::move(*NewBlockInfo);
         break;
       }
 
       BlockInfoCursor.SkipBlock();
     }
-
-    StreamFile.takeBlockInfo(std::move(BlockInfoFile));
   }
 
   unsigned NumTopBlocks = 0;
 
   // Parse the top-level structure.  We only allow blocks at the top-level.
   while (!Stream.AtEndOfStream()) {
     unsigned Code = Stream.ReadCode();
     if (Code != bitc::ENTER_SUBBLOCK)
       return ReportError("Invalid record at top-level");
 
     unsigned BlockID = Stream.ReadSubBlockID();
 
-    if (ParseBlock(Stream, BlockID, 0, CurStreamType))
+    if (ParseBlock(Stream, BlockInfo, BlockID, 0, CurStreamType))
       return true;
     ++NumTopBlocks;
   }
 
   if (Dump) outs() << "\n\n";
 
-  uint64_t BufferSizeBits = StreamFile.getBitcodeBytes().size() * CHAR_BIT;
+  uint64_t BufferSizeBits = Stream.getBitcodeBytes().size() * CHAR_BIT;
   // Print a summary of the read file.
   outs() << "Summary of " << InputFilename << ":\n";
   outs() << "         Total size: ";
   PrintSize(BufferSizeBits);
   outs() << "\n";
   outs() << "        Stream type: ";
   switch (CurStreamType) {
   case UnknownBitstream: outs() << "unknown\n"; break;
   case LLVMIRBitstream:  outs() << "LLVM IR\n"; break;
   }
   outs() << "  # Toplevel Blocks: " << NumTopBlocks << "\n";
   outs() << "\n";
 
   // Emit per-block stats.
   outs() << "Per-block Summary:\n";
   for (std::map<unsigned, PerBlockIDStats>::iterator I = BlockIDStats.begin(),
        E = BlockIDStats.end(); I != E; ++I) {
     outs() << "  Block ID #" << I->first;
-    if (const char *BlockName = GetBlockName(I->first, StreamFile,
-                                             CurStreamType))
+    if (const char *BlockName =
+            GetBlockName(I->first, BlockInfo, CurStreamType))
       outs() << " (" << BlockName << ")";
     outs() << ":\n";
 
     const PerBlockIDStats &Stats = I->second;
     outs() << "      Num Instances: " << Stats.NumInstances << "\n";
     outs() << "         Total Size: ";
     PrintSize(Stats.NumBits);
     outs() << "\n";
if (!NoHistogram && !Stats.CodeFreq.empty()) {
         if (RecStats.NumAbbrev)
           outs() <<
               format(" %7.2f",
                      (double)RecStats.NumAbbrev/RecStats.NumInstances*100);
         else
           outs() << "        ";
 
         outs() << "  ";
-        if (const char *CodeName =
-              GetCodeName(FreqPairs[i].second, I->first, StreamFile,
-                          CurStreamType))
+        if (const char *CodeName = GetCodeName(FreqPairs[i].second, I->first,
+                                               BlockInfo, CurStreamType))
           outs() << CodeName << "\n";
         else
           outs() << "UnknownCode" << FreqPairs[i].second << "\n";
       }
       outs() << "\n";
 
     }
   }

llvm/unittests/Bitcode/BitstreamReaderTest.cpp

 using namespace llvm;
 
 namespace {
 
 TEST(BitstreamReaderTest, AtEndOfStream) {
   uint8_t Bytes[4] = {
     0x00, 0x01, 0x02, 0x03
   };
-  BitstreamReader Reader(Bytes);
-  BitstreamCursor Cursor(Reader);
+  BitstreamCursor Cursor(Bytes);
 
   EXPECT_FALSE(Cursor.AtEndOfStream());
   (void)Cursor.Read(8);
   EXPECT_FALSE(Cursor.AtEndOfStream());
   (void)Cursor.Read(24);
   EXPECT_TRUE(Cursor.AtEndOfStream());
 
   Cursor.JumpToBit(0);
   EXPECT_FALSE(Cursor.AtEndOfStream());
 
   Cursor.JumpToBit(32);
   EXPECT_TRUE(Cursor.AtEndOfStream());
 }
 
 TEST(BitstreamReaderTest, AtEndOfStreamJump) {
   uint8_t Bytes[4] = {
     0x00, 0x01, 0x02, 0x03
   };
-  BitstreamReader Reader(Bytes);
-  BitstreamCursor Cursor(Reader);
+  BitstreamCursor Cursor(Bytes);
 
   Cursor.JumpToBit(32);
   EXPECT_TRUE(Cursor.AtEndOfStream());
 }
 
 TEST(BitstreamReaderTest, AtEndOfStreamEmpty) {
-  BitstreamReader Reader(ArrayRef<uint8_t>{});
-  BitstreamCursor Cursor(Reader);
+  BitstreamCursor Cursor(ArrayRef<uint8_t>{});
 
   EXPECT_TRUE(Cursor.AtEndOfStream());
 }
 
 TEST(BitstreamReaderTest, getCurrentByteNo) {
   uint8_t Bytes[] = {0x00, 0x01, 0x02, 0x03};
-  BitstreamReader Reader(Bytes);
-  SimpleBitstreamCursor Cursor(Reader);
+  SimpleBitstreamCursor Cursor(Bytes);
 
   for (unsigned I = 0, E = 32; I != E; ++I) {
     EXPECT_EQ(I / 8, Cursor.getCurrentByteNo());
     (void)Cursor.Read(1);
   }
   EXPECT_EQ(4u, Cursor.getCurrentByteNo());
 }
 
 TEST(BitstreamReaderTest, getPointerToByte) {
   uint8_t Bytes[] = {0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07};
-  BitstreamReader Reader(Bytes);
-  SimpleBitstreamCursor Cursor(Reader);
+  SimpleBitstreamCursor Cursor(Bytes);
 
   for (unsigned I = 0, E = 8; I != E; ++I) {
     EXPECT_EQ(Bytes + I, Cursor.getPointerToByte(I, 1));
   }
 }
 
 TEST(BitstreamReaderTest, getPointerToBit) {
   uint8_t Bytes[] = {0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07};
-  BitstreamReader Reader(Bytes);
-  SimpleBitstreamCursor Cursor(Reader);
+  SimpleBitstreamCursor Cursor(Bytes);
 
   for (unsigned I = 0, E = 8; I != E; ++I) {
     EXPECT_EQ(Bytes + I, Cursor.getPointerToBit(I * 8, 1));
   }
 }
 
-TEST(BitstreamReaderTest, jumpToPointer) {
-  uint8_t Bytes[] = {0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07};
-  BitstreamReader Reader(Bytes);
-  SimpleBitstreamCursor Cursor(Reader);
-
-  for (unsigned I : {0, 6, 2, 7}) {
-    Cursor.jumpToPointer(Bytes + I);
-    EXPECT_EQ(I, Cursor.getCurrentByteNo());
-  }
-}
-
 TEST(BitstreamReaderTest, readRecordWithBlobWhileStreaming) {
   SmallVector<uint8_t, 1> BlobData;
   for (unsigned I = 0, E = 1024; I != E; ++I)
     BlobData.push_back(I);
 
   // Try a bunch of different sizes.
   const unsigned Magic = 0x12345678;
   const unsigned BlockID = bitc::FIRST_APPLICATION_BLOCKID;
unsigned AbbrevID;
       AbbrevID = Stream.EmitAbbrev(Abbrev);
       unsigned Record[] = {RecordID};
       Stream.EmitRecordWithBlob(AbbrevID, makeArrayRef(Record), BlobIn);
 
       Stream.ExitBlock();
     }
 
     // Stream the buffer into the reader.
-    BitstreamReader R(
+    BitstreamCursor Stream(
         ArrayRef<uint8_t>((const uint8_t *)Buffer.begin(), Buffer.size()));
-    BitstreamCursor Stream(R);
 
     // Header.  Included in test so that we can run llvm-bcanalyzer to debug
     // when there are problems.
     ASSERT_EQ(Magic, Stream.Read(32));
 
     // Block.
     BitstreamEntry Entry =
         Stream.advance(BitstreamCursor::AF_DontAutoprocessAbbrevs);
for (unsigned I = 0, BlobSize = 0, E = BlobData.size(); BlobSize < E;
     EXPECT_TRUE(Record.empty());
     EXPECT_EQ(BlobIn, BlobOut);
   }
 }
 
 TEST(BitstreamReaderTest, shortRead) {
   uint8_t Bytes[] = {8, 7, 6, 5, 4, 3, 2, 1};
   for (unsigned I = 1; I != 8; ++I) {
-    BitstreamReader Reader(ArrayRef<uint8_t>(Bytes, I));
-    SimpleBitstreamCursor Cursor(Reader);
+    SimpleBitstreamCursor Cursor(ArrayRef<uint8_t>(Bytes, I));
     EXPECT_EQ(8ull, Cursor.Read(8));
   }
 }
 
 } // end anonymous namespace