BitcodeReader.h revision d724d097437f40a5689464429f948ec41e4a2415 
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 BITCODE_READER_H
15#define BITCODE_READER_H
16
17#include "llvm/ADT/DenseMap.h"
18#include "llvm/Bitcode/BitstreamReader.h"
19#include "llvm/Bitcode/LLVMBitCodes.h"
20#include "llvm/GVMaterializer.h"
21#include "llvm/IR/Attributes.h"
22#include "llvm/IR/OperandTraits.h"
23#include "llvm/IR/Type.h"
24#include "llvm/Support/ValueHandle.h"
25#include <vector>
26
27namespace llvm {
28  class MemoryBuffer;
29  class LLVMContext;
30}
31
32namespace llvm_2_7 {
33
34using namespace llvm;
35
36//===----------------------------------------------------------------------===//
37//                          BitcodeReaderValueList Class
38//===----------------------------------------------------------------------===//
39
40class 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;
53public:
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
99class BitcodeReaderMDValueList {
100  std::vector<WeakVH> MDValuePtrs;
101
102  LLVMContext &Context;
103public:
104  BitcodeReaderMDValueList(LLVMContext& C) : Context(C) {}
105
106  // vector compatibility methods
107  unsigned size() const       { return MDValuePtrs.size(); }
108  void resize(unsigned N)     { MDValuePtrs.resize(N); }
109  void push_back(Value *V)    { MDValuePtrs.push_back(V);  }
110  void clear()                { MDValuePtrs.clear();  }
111  Value *back() const         { return MDValuePtrs.back(); }
112  void pop_back()             { MDValuePtrs.pop_back(); }
113  bool empty() const          { return MDValuePtrs.empty(); }
114
115  Value *operator[](unsigned i) const {
116    assert(i < MDValuePtrs.size());
117    return MDValuePtrs[i];
118  }
119
120  void shrinkTo(unsigned N) {
121    assert(N <= size() && "Invalid shrinkTo request!");
122    MDValuePtrs.resize(N);
123  }
124
125  Value *getValueFwdRef(unsigned Idx);
126  void AssignValue(Value *V, unsigned Idx);
127};
128
129class BitcodeReader : public GVMaterializer {
130  LLVMContext &Context;
131  Module *TheModule;
132  MemoryBuffer *Buffer;
133  bool BufferOwned;
134  OwningPtr<BitstreamReader> StreamFile;
135  BitstreamCursor Stream;
136  DataStreamer *LazyStreamer;
137  uint64_t NextUnreadBit;
138  bool SeenValueSymbolTable;
139
140  std::vector<Type*> TypeList;
141  BitcodeReaderValueList ValueList;
142  BitcodeReaderMDValueList MDValueList;
143  SmallVector<Instruction *, 64> InstructionList;
144
145  std::vector<std::pair<GlobalVariable*, unsigned> > GlobalInits;
146  std::vector<std::pair<GlobalAlias*, unsigned> > AliasInits;
147
148  /// MAttributes - The set of attributes by index.  Index zero in the
149  /// file is for null, and is thus not represented here.  As such all indices
150  /// are off by one.
151  std::vector<AttributeSet> MAttributes;
152
153  /// \brief The set of attribute groups.
154  std::map<unsigned, AttributeSet> MAttributeGroups;
155
156  /// FunctionBBs - While parsing a function body, this is a list of the basic
157  /// blocks for the function.
158  std::vector<BasicBlock*> FunctionBBs;
159
160  // When reading the module header, this list is populated with functions that
161  // have bodies later in the file.
162  std::vector<Function*> FunctionsWithBodies;
163
164  // When intrinsic functions are encountered which require upgrading they are
165  // stored here with their replacement function.
166  typedef std::vector<std::pair<Function*, Function*> > UpgradedIntrinsicMap;
167  UpgradedIntrinsicMap UpgradedIntrinsics;
168
169  // Map the bitcode's custom MDKind ID to the Module's MDKind ID.
170  DenseMap<unsigned, unsigned> MDKindMap;
171
172  // Several operations happen after the module header has been read, but
173  // before function bodies are processed. This keeps track of whether
174  // we've done this yet.
175  bool SeenFirstFunctionBody;
176
177  /// DeferredFunctionInfo - When function bodies are initially scanned, this
178  /// map contains info about where to find deferred function body in the
179  /// stream.
180  DenseMap<Function*, uint64_t> DeferredFunctionInfo;
181
182  /// BlockAddrFwdRefs - These are blockaddr references to basic blocks.  These
183  /// are resolved lazily when functions are loaded.
184  typedef std::pair<unsigned, GlobalVariable*> BlockAddrRefTy;
185  DenseMap<Function*, std::vector<BlockAddrRefTy> > BlockAddrFwdRefs;
186
187  /// LLVM2_7MetadataDetected - True if metadata produced by LLVM 2.7 or
188  /// earlier was detected, in which case we behave slightly differently,
189  /// for compatibility.
190  /// FIXME: Remove in LLVM 3.0.
191  bool LLVM2_7MetadataDetected;
192  static const error_category &BitcodeErrorCategory();
193
194public:
195  enum ErrorType {
196    BitcodeStreamInvalidSize,
197    ConflictingMETADATA_KINDRecords,
198    CouldNotFindFunctionInStream,
199    ExpectedConstant,
200    InsufficientFunctionProtos,
201    InvalidBitcodeSignature,
202    InvalidBitcodeWrapperHeader,
203    InvalidConstantReference,
204    InvalidID, // A read identifier is not found in the table it should be in.
205    InvalidInstructionWithNoBB,
206    InvalidRecord, // A read record doesn't have the expected size or structure
207    InvalidTypeForValue, // Type read OK, but is invalid for its use
208    InvalidTYPETable,
209    InvalidType, // We were unable to read a type
210    MalformedBlock, // We are unable to advance in the stream.
211    MalformedGlobalInitializerSet,
212    InvalidMultipleBlocks, // We found multiple blocks of a kind that should
213                           // have only one
214    NeverResolvedValueFoundInFunction,
215    InvalidValue // Invalid version, inst number, attr number, etc
216  };
217
218  error_code Error(ErrorType E) {
219    return error_code(E, BitcodeErrorCategory());
220  }
221
222  explicit BitcodeReader(MemoryBuffer *buffer, LLVMContext &C)
223    : Context(C), TheModule(0), Buffer(buffer), BufferOwned(false),
224      LazyStreamer(0), NextUnreadBit(0), SeenValueSymbolTable(false),
225      ValueList(C), MDValueList(C),
226      SeenFirstFunctionBody(false), LLVM2_7MetadataDetected(false) {
227  }
228  ~BitcodeReader() {
229    FreeState();
230  }
231
232  void FreeState();
233
234  /// setBufferOwned - If this is true, the reader will destroy the MemoryBuffer
235  /// when the reader is destroyed.
236  void setBufferOwned(bool Owned) { BufferOwned = Owned; }
237
238  virtual bool isMaterializable(const GlobalValue *GV) const;
239  virtual bool isDematerializable(const GlobalValue *GV) const;
240  virtual error_code Materialize(GlobalValue *GV);
241  virtual error_code MaterializeModule(Module *M);
242  virtual void Dematerialize(GlobalValue *GV);
243
244  /// @brief Main interface to parsing a bitcode buffer.
245  /// @returns true if an error occurred.
246  error_code ParseBitcodeInto(Module *M);
247
248  /// @brief Cheap mechanism to just extract module triple
249  /// @returns true if an error occurred.
250  error_code ParseTriple(std::string &Triple);
251
252  static uint64_t decodeSignRotatedValue(uint64_t V);
253
254private:
255  Type *getTypeByID(unsigned ID);
256  Type *getTypeByIDOrNull(unsigned ID);
257  Value *getFnValueByID(unsigned ID, Type *Ty) {
258    if (Ty && Ty->isMetadataTy())
259      return MDValueList.getValueFwdRef(ID);
260    return ValueList.getValueFwdRef(ID, Ty);
261  }
262  BasicBlock *getBasicBlock(unsigned ID) const {
263    if (ID >= FunctionBBs.size()) return 0; // Invalid ID
264    return FunctionBBs[ID];
265  }
266  AttributeSet getAttributes(unsigned i) const {
267    if (i-1 < MAttributes.size())
268      return MAttributes[i-1];
269    return AttributeSet();
270  }
271
272  /// getValueTypePair - Read a value/type pair out of the specified record from
273  /// slot 'Slot'.  Increment Slot past the number of slots used in the record.
274  /// Return true on failure.
275  bool getValueTypePair(SmallVector<uint64_t, 64> &Record, unsigned &Slot,
276                        unsigned InstNum, Value *&ResVal) {
277    if (Slot == Record.size()) return true;
278    unsigned ValNo = (unsigned)Record[Slot++];
279    if (ValNo < InstNum) {
280      // If this is not a forward reference, just return the value we already
281      // have.
282      ResVal = getFnValueByID(ValNo, 0);
283      return ResVal == 0;
284    } else if (Slot == Record.size()) {
285      return true;
286    }
287
288    unsigned TypeNo = (unsigned)Record[Slot++];
289    ResVal = getFnValueByID(ValNo, getTypeByID(TypeNo));
290    return ResVal == 0;
291  }
292  bool getValue(SmallVector<uint64_t, 64> &Record, unsigned &Slot,
293                Type *Ty, Value *&ResVal) {
294    if (Slot == Record.size()) return true;
295    unsigned ValNo = (unsigned)Record[Slot++];
296    ResVal = getFnValueByID(ValNo, Ty);
297    return ResVal == 0;
298  }
299
300
301  error_code ParseModule(bool Resume);
302  error_code ParseAttributeBlock();
303  error_code ParseTypeTable();
304  error_code ParseOldTypeTable();         // FIXME: Remove in LLVM 3.1
305  error_code ParseTypeTableBody();
306
307  error_code ParseOldTypeSymbolTable();   // FIXME: Remove in LLVM 3.1
308  error_code ParseValueSymbolTable();
309  error_code ParseConstants();
310  error_code RememberAndSkipFunctionBody();
311  error_code ParseFunctionBody(Function *F);
312  error_code GlobalCleanup();
313  error_code ResolveGlobalAndAliasInits();
314  error_code ParseMetadata();
315  error_code ParseMetadataAttachment();
316  error_code ParseModuleTriple(std::string &Triple);
317  error_code InitStream();
318  error_code InitStreamFromBuffer();
319  error_code InitLazyStream();
320};
321
322} // End llvm_2_7 namespace
323
324#endif
325