src/llvm/lib/Bitcode/Reader/BitcodeReader.h

   1 //===- BitcodeReader.h - Internal BitcodeReader impl ------------*- C++ -*-===//
   2 //
   3 //                     The LLVM Compiler Infrastructure
   4 //
   5 // This file is distributed under the University of Illinois Open Source
   6 // License. See LICENSE.TXT for details.
   7 //
   8 //===----------------------------------------------------------------------===//
   9 //
  10 // This header defines the BitcodeReader class.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #ifndef LLVM_LIB_BITCODE_READER_BITCODEREADER_H
  15 #define LLVM_LIB_BITCODE_READER_BITCODEREADER_H
  16
  17 #include "llvm/ADT/DenseMap.h"
  18 #include "llvm/Bitcode/BitstreamReader.h"
  19 #include "llvm/Bitcode/LLVMBitCodes.h"
  20 #include "llvm/IR/Attributes.h"
  21 #include "llvm/IR/GVMaterializer.h"
  22 #include "llvm/IR/Metadata.h"
  23 #include "llvm/IR/OperandTraits.h"
  24 #include "llvm/IR/TrackingMDRef.h"
  25 #include "llvm/IR/Type.h"
  26 #include "llvm/IR/ValueHandle.h"
  27 #include <deque>
  28 #include <system_error>
  29 #include <vector>
  30
  31 namespace llvm {
  32   class Comdat;
  33   class MemoryBuffer;
  34   class LLVMContext;
  35
  36 //===----------------------------------------------------------------------===//
  37 //                          BitcodeReaderValueList Class
  38 //===----------------------------------------------------------------------===//
  39
  40 class BitcodeReaderValueList {
  41   std::vector<WeakVH> ValuePtrs;
  42
  43   /// ResolveConstants - As we resolve forward-referenced constants, we add
  44   /// information about them to this vector.  This allows us to resolve them in
  45   /// bulk instead of resolving each reference at a time.  See the code in
  46   /// ResolveConstantForwardRefs for more information about this.
  47   ///
  48   /// The key of this vector is the placeholder constant, the value is the slot
  49   /// number that holds the resolved value.
  50   typedef std::vector<std::pair<Constant*, unsigned> > ResolveConstantsTy;
  51   ResolveConstantsTy ResolveConstants;
  52   LLVMContext &Context;
  53 public:
  54   BitcodeReaderValueList(LLVMContext &C) : Context(C) {}
  55   ~BitcodeReaderValueList() {
  56     assert(ResolveConstants.empty() && "Constants not resolved?");
  57   }
  58
  59   // vector compatibility methods
  60   unsigned size() const { return ValuePtrs.size(); }
  61   void resize(unsigned N) { ValuePtrs.resize(N); }
  62   void push_back(Value *V) {
  63     ValuePtrs.push_back(V);
  64   }
  65
  66   void clear() {
  67     assert(ResolveConstants.empty() && "Constants not resolved?");
  68     ValuePtrs.clear();
  69   }
  70
  71   Value *operator[](unsigned i) const {
  72     assert(i < ValuePtrs.size());
  73     return ValuePtrs[i];
  74   }
  75
  76   Value *back() const { return ValuePtrs.back(); }
  77     void pop_back() { ValuePtrs.pop_back(); }
  78   bool empty() const { return ValuePtrs.empty(); }
  79   void shrinkTo(unsigned N) {
  80     assert(N <= size() && "Invalid shrinkTo request!");
  81     ValuePtrs.resize(N);
  82   }
  83
  84   Constant *getConstantFwdRef(unsigned Idx, Type *Ty);
  85   Value *getValueFwdRef(unsigned Idx, Type *Ty);
  86
  87   void AssignValue(Value *V, unsigned Idx);
  88
  89   /// ResolveConstantForwardRefs - Once all constants are read, this method bulk
  90   /// resolves any forward references.
  91   void ResolveConstantForwardRefs();
  92 };
  93
  94
  95 //===----------------------------------------------------------------------===//
  96 //                          BitcodeReaderMDValueList Class
  97 //===----------------------------------------------------------------------===//
  98
  99 class BitcodeReaderMDValueList {
 100   unsigned NumFwdRefs;
 101   bool AnyFwdRefs;
 102   std::vector<TrackingMDRef> MDValuePtrs;
 103
 104   LLVMContext &Context;
 105 public:
 106   BitcodeReaderMDValueList(LLVMContext &C)
 107       : NumFwdRefs(0), AnyFwdRefs(false), Context(C) {}
 108
 109   // vector compatibility methods
 110   unsigned size() const       { return MDValuePtrs.size(); }
 111   void resize(unsigned N)     { MDValuePtrs.resize(N); }
 112   void push_back(Metadata *MD) { MDValuePtrs.emplace_back(MD); }
 113   void clear()                { MDValuePtrs.clear();  }
 114   Metadata *back() const      { return MDValuePtrs.back(); }
 115   void pop_back()             { MDValuePtrs.pop_back(); }
 116   bool empty() const          { return MDValuePtrs.empty(); }
 117
 118   Metadata *operator[](unsigned i) const {
 119     assert(i < MDValuePtrs.size());
 120     return MDValuePtrs[i];
 121   }
 122
 123   void shrinkTo(unsigned N) {
 124     assert(N <= size() && "Invalid shrinkTo request!");
 125     MDValuePtrs.resize(N);
 126   }
 127
 128   Metadata *getValueFwdRef(unsigned Idx);
 129   void AssignValue(Metadata *MD, unsigned Idx);
 130   void tryToResolveCycles();
 131 };
 132
 133 class BitcodeReader : public GVMaterializer {
 134   LLVMContext &Context;
 135   DiagnosticHandlerFunction DiagnosticHandler;
 136   Module *TheModule;
 137   std::unique_ptr<MemoryBuffer> Buffer;
 138   std::unique_ptr<BitstreamReader> StreamFile;
 139   BitstreamCursor Stream;
 140   DataStreamer *LazyStreamer;
 141   uint64_t NextUnreadBit;
 142   bool SeenValueSymbolTable;
 143
 144   std::vector<Type*> TypeList;
 145   BitcodeReaderValueList ValueList;
 146   BitcodeReaderMDValueList MDValueList;
 147   std::vector<Comdat *> ComdatList;
 148   SmallVector<Instruction *, 64> InstructionList;
 149
 150   std::vector<std::pair<GlobalVariable*, unsigned> > GlobalInits;
 151   std::vector<std::pair<GlobalAlias*, unsigned> > AliasInits;
 152   std::vector<std::pair<Function*, unsigned> > FunctionPrefixes;
 153   std::vector<std::pair<Function*, unsigned> > FunctionPrologues;
 154
 155   SmallVector<Instruction*, 64> InstsWithTBAATag;
 156
 157   /// MAttributes - The set of attributes by index.  Index zero in the
 158   /// file is for null, and is thus not represented here.  As such all indices
 159   /// are off by one.
 160   std::vector<AttributeSet> MAttributes;
 161
 162   /// \brief The set of attribute groups.
 163   std::map<unsigned, AttributeSet> MAttributeGroups;
 164
 165   /// FunctionBBs - While parsing a function body, this is a list of the basic
 166   /// blocks for the function.
 167   std::vector<BasicBlock*> FunctionBBs;
 168
 169   // When reading the module header, this list is populated with functions that
 170   // have bodies later in the file.
 171   std::vector<Function*> FunctionsWithBodies;
 172
 173   // When intrinsic functions are encountered which require upgrading they are
 174   // stored here with their replacement function.
 175   typedef std::vector<std::pair<Function*, Function*> > UpgradedIntrinsicMap;
 176   UpgradedIntrinsicMap UpgradedIntrinsics;
 177
 178   // Map the bitcode's custom MDKind ID to the Module's MDKind ID.
 179   DenseMap<unsigned, unsigned> MDKindMap;
 180
 181   // Several operations happen after the module header has been read, but
 182   // before function bodies are processed. This keeps track of whether
 183   // we've done this yet.
 184   bool SeenFirstFunctionBody;
 185
 186   /// DeferredFunctionInfo - When function bodies are initially scanned, this
 187   /// map contains info about where to find deferred function body in the
 188   /// stream.
 189   DenseMap<Function*, uint64_t> DeferredFunctionInfo;
 190
 191   /// These are basic blocks forward-referenced by block addresses.  They are
 192   /// inserted lazily into functions when they're loaded.  The basic block ID is
 193   /// its index into the vector.
 194   DenseMap<Function *, std::vector<BasicBlock *>> BasicBlockFwdRefs;
 195   std::deque<Function *> BasicBlockFwdRefQueue;
 196
 197   /// UseRelativeIDs - Indicates that we are using a new encoding for
 198   /// instruction operands where most operands in the current
 199   /// FUNCTION_BLOCK are encoded relative to the instruction number,
 200   /// for a more compact encoding.  Some instruction operands are not
 201   /// relative to the instruction ID: basic block numbers, and types.
 202   /// Once the old style function blocks have been phased out, we would
 203   /// not need this flag.
 204   bool UseRelativeIDs;
 205
 206   /// True if all functions will be materialized, negating the need to process
 207   /// (e.g.) blockaddress forward references.
 208   bool WillMaterializeAllForwardRefs;
 209
 210   /// Functions that have block addresses taken.  This is usually empty.
 211   SmallPtrSet<const Function *, 4> BlockAddressesTaken;
 212
 213 public:
 214   std::error_code Error(BitcodeError E, const Twine &Message);
 215   std::error_code Error(BitcodeError E);
 216   std::error_code Error(const Twine &Message);
 217
 218   explicit BitcodeReader(MemoryBuffer *buffer, LLVMContext &C,
 219                          DiagnosticHandlerFunction DiagnosticHandler);
 220   explicit BitcodeReader(DataStreamer *streamer, LLVMContext &C,
 221                          DiagnosticHandlerFunction DiagnosticHandler);
 222   ~BitcodeReader() { FreeState(); }
 223
 224   std::error_code materializeForwardReferencedFunctions();
 225
 226   void FreeState();
 227
 228   void releaseBuffer();
 229
 230   bool isDematerializable(const GlobalValue *GV) const override;
 231   std::error_code materialize(GlobalValue *GV) override;
 232   std::error_code MaterializeModule(Module *M) override;
 233   std::vector<StructType *> getIdentifiedStructTypes() const override;
 234   void Dematerialize(GlobalValue *GV) override;
 235
 236   /// @brief Main interface to parsing a bitcode buffer.
 237   /// @returns true if an error occurred.
 238   std::error_code ParseBitcodeInto(Module *M);
 239
 240   /// @brief Cheap mechanism to just extract module triple
 241   /// @returns true if an error occurred.
 242   ErrorOr<std::string> parseTriple();
 243
 244   static uint64_t decodeSignRotatedValue(uint64_t V);
 245
 246 private:
 247   std::vector<StructType *> IdentifiedStructTypes;
 248   StructType *createIdentifiedStructType(LLVMContext &Context, StringRef Name);
 249   StructType *createIdentifiedStructType(LLVMContext &Context);
 250
 251   Type *getTypeByID(unsigned ID);
 252   Value *getFnValueByID(unsigned ID, Type *Ty) {
 253     if (Ty && Ty->isMetadataTy())
 254       return MetadataAsValue::get(Ty->getContext(), getFnMetadataByID(ID));
 255     return ValueList.getValueFwdRef(ID, Ty);
 256   }
 257   Metadata *getFnMetadataByID(unsigned ID) {
 258     return MDValueList.getValueFwdRef(ID);
 259   }
 260   BasicBlock *getBasicBlock(unsigned ID) const {
 261     if (ID >= FunctionBBs.size()) return nullptr; // Invalid ID
 262     return FunctionBBs[ID];
 263   }
 264   AttributeSet getAttributes(unsigned i) const {
 265     if (i-1 < MAttributes.size())
 266       return MAttributes[i-1];
 267     return AttributeSet();
 268   }
 269
 270   /// getValueTypePair - Read a value/type pair out of the specified record from
 271   /// slot 'Slot'.  Increment Slot past the number of slots used in the record.
 272   /// Return true on failure.
 273   bool getValueTypePair(SmallVectorImpl<uint64_t> &Record, unsigned &Slot,
 274                         unsigned InstNum, Value *&ResVal) {
 275     if (Slot == Record.size()) return true;
 276     unsigned ValNo = (unsigned)Record[Slot++];
 277     // Adjust the ValNo, if it was encoded relative to the InstNum.
 278     if (UseRelativeIDs)
 279       ValNo = InstNum - ValNo;
 280     if (ValNo < InstNum) {
 281       // If this is not a forward reference, just return the value we already
 282       // have.
 283       ResVal = getFnValueByID(ValNo, nullptr);
 284       return ResVal == nullptr;
 285     } else if (Slot == Record.size()) {
 286       return true;
 287     }
 288
 289     unsigned TypeNo = (unsigned)Record[Slot++];
 290     ResVal = getFnValueByID(ValNo, getTypeByID(TypeNo));
 291     return ResVal == nullptr;
 292   }
 293
 294   /// popValue - Read a value out of the specified record from slot 'Slot'.
 295   /// Increment Slot past the number of slots used by the value in the record.
 296   /// Return true if there is an error.
 297   bool popValue(SmallVectorImpl<uint64_t> &Record, unsigned &Slot,
 298                 unsigned InstNum, Type *Ty, Value *&ResVal) {
 299     if (getValue(Record, Slot, InstNum, Ty, ResVal))
 300       return true;
 301     // All values currently take a single record slot.
 302     ++Slot;
 303     return false;
 304   }
 305
 306   /// getValue -- Like popValue, but does not increment the Slot number.
 307   bool getValue(SmallVectorImpl<uint64_t> &Record, unsigned Slot,
 308                 unsigned InstNum, Type *Ty, Value *&ResVal) {
 309     ResVal = getValue(Record, Slot, InstNum, Ty);
 310     return ResVal == nullptr;
 311   }
 312
 313   /// getValue -- Version of getValue that returns ResVal directly,
 314   /// or 0 if there is an error.
 315   Value *getValue(SmallVectorImpl<uint64_t> &Record, unsigned Slot,
 316                   unsigned InstNum, Type *Ty) {
 317     if (Slot == Record.size()) return nullptr;
 318     unsigned ValNo = (unsigned)Record[Slot];
 319     // Adjust the ValNo, if it was encoded relative to the InstNum.
 320     if (UseRelativeIDs)
 321       ValNo = InstNum - ValNo;
 322     return getFnValueByID(ValNo, Ty);
 323   }
 324
 325   /// getValueSigned -- Like getValue, but decodes signed VBRs.
 326   Value *getValueSigned(SmallVectorImpl<uint64_t> &Record, unsigned Slot,
 327                         unsigned InstNum, Type *Ty) {
 328     if (Slot == Record.size()) return nullptr;
 329     unsigned ValNo = (unsigned)decodeSignRotatedValue(Record[Slot]);
 330     // Adjust the ValNo, if it was encoded relative to the InstNum.
 331     if (UseRelativeIDs)
 332       ValNo = InstNum - ValNo;
 333     return getFnValueByID(ValNo, Ty);
 334   }
 335
 336   std::error_code ParseAttrKind(uint64_t Code, Attribute::AttrKind *Kind);
 337   std::error_code ParseModule(bool Resume);
 338   std::error_code ParseAttributeBlock();
 339   std::error_code ParseAttributeGroupBlock();
 340   std::error_code ParseTypeTable();
 341   std::error_code ParseTypeTableBody();
 342
 343   std::error_code ParseValueSymbolTable();
 344   std::error_code ParseConstants();
 345   std::error_code RememberAndSkipFunctionBody();
 346   std::error_code ParseFunctionBody(Function *F);
 347   std::error_code GlobalCleanup();
 348   std::error_code ResolveGlobalAndAliasInits();
 349   std::error_code ParseMetadata();
 350   std::error_code ParseMetadataAttachment();
 351   ErrorOr<std::string> parseModuleTriple();
 352   std::error_code ParseUseLists();
 353   std::error_code InitStream();
 354   std::error_code InitStreamFromBuffer();
 355   std::error_code InitLazyStream();
 356   std::error_code FindFunctionInStream(
 357       Function *F,
 358       DenseMap<Function *, uint64_t>::iterator DeferredFunctionInfoIterator);
 359 };
 360
 361 } // End llvm namespace
 362
 363 #endif