1//===-- llvm/MC/MCMachObjectWriter.h - Mach Object Writer -------*- 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#ifndef LLVM_MC_MCMACHOBJECTWRITER_H
11#define LLVM_MC_MCMACHOBJECTWRITER_H
12
13#include "llvm/ADT/DenseMap.h"
14#include "llvm/ADT/OwningPtr.h"
15#include "llvm/ADT/SmallString.h"
16#include "llvm/MC/MCExpr.h"
17#include "llvm/MC/MCObjectWriter.h"
18#include "llvm/Object/MachOFormat.h"
19#include "llvm/Support/DataTypes.h"
20#include <vector>
21
22namespace llvm {
23
24class MCSectionData;
25class MachObjectWriter;
26
27class MCMachObjectTargetWriter {
28  const unsigned Is64Bit : 1;
29  const uint32_t CPUType;
30  const uint32_t CPUSubtype;
31  // FIXME: Remove this, we should just always use it once we no longer care
32  // about Darwin 'as' compatibility.
33  const unsigned UseAggressiveSymbolFolding : 1;
34  unsigned LocalDifference_RIT;
35
36protected:
37  MCMachObjectTargetWriter(bool Is64Bit_, uint32_t CPUType_,
38                           uint32_t CPUSubtype_,
39                           bool UseAggressiveSymbolFolding_ = false);
40
41  void setLocalDifferenceRelocationType(unsigned Type) {
42    LocalDifference_RIT = Type;
43  }
44
45public:
46  virtual ~MCMachObjectTargetWriter();
47
48  /// @name Accessors
49  /// @{
50
51  bool is64Bit() const { return Is64Bit; }
52  bool useAggressiveSymbolFolding() const { return UseAggressiveSymbolFolding; }
53  uint32_t getCPUType() const { return CPUType; }
54  uint32_t getCPUSubtype() const { return CPUSubtype; }
55  unsigned getLocalDifferenceRelocationType() const {
56    return LocalDifference_RIT;
57  }
58
59  /// @}
60
61  /// @name API
62  /// @{
63
64  virtual void RecordRelocation(MachObjectWriter *Writer,
65                                const MCAssembler &Asm,
66                                const MCAsmLayout &Layout,
67                                const MCFragment *Fragment,
68                                const MCFixup &Fixup,
69                                MCValue Target,
70                                uint64_t &FixedValue) = 0;
71
72  /// @}
73};
74
75class MachObjectWriter : public MCObjectWriter {
76  /// MachSymbolData - Helper struct for containing some precomputed information
77  /// on symbols.
78  struct MachSymbolData {
79    MCSymbolData *SymbolData;
80    uint64_t StringIndex;
81    uint8_t SectionIndex;
82
83    // Support lexicographic sorting.
84    bool operator<(const MachSymbolData &RHS) const;
85  };
86
87  /// The target specific Mach-O writer instance.
88  llvm::OwningPtr<MCMachObjectTargetWriter> TargetObjectWriter;
89
90  /// @name Relocation Data
91  /// @{
92
93  llvm::DenseMap<const MCSectionData*,
94                 std::vector<object::macho::RelocationEntry> > Relocations;
95  llvm::DenseMap<const MCSectionData*, unsigned> IndirectSymBase;
96
97  /// @}
98  /// @name Symbol Table Data
99  /// @{
100
101  SmallString<256> StringTable;
102  std::vector<MachSymbolData> LocalSymbolData;
103  std::vector<MachSymbolData> ExternalSymbolData;
104  std::vector<MachSymbolData> UndefinedSymbolData;
105
106  /// @}
107
108public:
109  MachObjectWriter(MCMachObjectTargetWriter *MOTW, raw_ostream &_OS,
110                   bool _IsLittleEndian)
111    : MCObjectWriter(_OS, _IsLittleEndian), TargetObjectWriter(MOTW) {
112  }
113
114  /// @name Utility Methods
115  /// @{
116
117  bool isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind);
118
119  SectionAddrMap SectionAddress;
120
121  SectionAddrMap &getSectionAddressMap() { return SectionAddress; }
122
123  uint64_t getSectionAddress(const MCSectionData* SD) const {
124    return SectionAddress.lookup(SD);
125  }
126  uint64_t getSymbolAddress(const MCSymbolData* SD,
127                            const MCAsmLayout &Layout) const;
128
129  uint64_t getFragmentAddress(const MCFragment *Fragment,
130                              const MCAsmLayout &Layout) const;
131
132  uint64_t getPaddingSize(const MCSectionData *SD,
133                          const MCAsmLayout &Layout) const;
134
135  bool doesSymbolRequireExternRelocation(const MCSymbolData *SD);
136
137  /// @}
138
139  /// @name Target Writer Proxy Accessors
140  /// @{
141
142  bool is64Bit() const { return TargetObjectWriter->is64Bit(); }
143  bool isARM() const {
144    uint32_t CPUType = TargetObjectWriter->getCPUType() &
145      ~object::mach::CTFM_ArchMask;
146    return CPUType == object::mach::CTM_ARM;
147  }
148
149  /// @}
150
151  void WriteHeader(unsigned NumLoadCommands, unsigned LoadCommandsSize,
152                   bool SubsectionsViaSymbols);
153
154  /// WriteSegmentLoadCommand - Write a segment load command.
155  ///
156  /// \arg NumSections - The number of sections in this segment.
157  /// \arg SectionDataSize - The total size of the sections.
158  void WriteSegmentLoadCommand(unsigned NumSections,
159                               uint64_t VMSize,
160                               uint64_t SectionDataStartOffset,
161                               uint64_t SectionDataSize);
162
163  void WriteSection(const MCAssembler &Asm, const MCAsmLayout &Layout,
164                    const MCSectionData &SD, uint64_t FileOffset,
165                    uint64_t RelocationsStart, unsigned NumRelocations);
166
167  void WriteSymtabLoadCommand(uint32_t SymbolOffset, uint32_t NumSymbols,
168                              uint32_t StringTableOffset,
169                              uint32_t StringTableSize);
170
171  void WriteDysymtabLoadCommand(uint32_t FirstLocalSymbol,
172                                uint32_t NumLocalSymbols,
173                                uint32_t FirstExternalSymbol,
174                                uint32_t NumExternalSymbols,
175                                uint32_t FirstUndefinedSymbol,
176                                uint32_t NumUndefinedSymbols,
177                                uint32_t IndirectSymbolOffset,
178                                uint32_t NumIndirectSymbols);
179
180  void WriteNlist(MachSymbolData &MSD, const MCAsmLayout &Layout);
181
182  // FIXME: We really need to improve the relocation validation. Basically, we
183  // want to implement a separate computation which evaluates the relocation
184  // entry as the linker would, and verifies that the resultant fixup value is
185  // exactly what the encoder wanted. This will catch several classes of
186  // problems:
187  //
188  //  - Relocation entry bugs, the two algorithms are unlikely to have the same
189  //    exact bug.
190  //
191  //  - Relaxation issues, where we forget to relax something.
192  //
193  //  - Input errors, where something cannot be correctly encoded. 'as' allows
194  //    these through in many cases.
195
196  void addRelocation(const MCSectionData *SD,
197                     object::macho::RelocationEntry &MRE) {
198    Relocations[SD].push_back(MRE);
199  }
200
201  void RecordScatteredRelocation(const MCAssembler &Asm,
202                                 const MCAsmLayout &Layout,
203                                 const MCFragment *Fragment,
204                                 const MCFixup &Fixup, MCValue Target,
205                                 unsigned Log2Size,
206                                 uint64_t &FixedValue);
207
208  void RecordTLVPRelocation(const MCAssembler &Asm,
209                            const MCAsmLayout &Layout,
210                            const MCFragment *Fragment,
211                            const MCFixup &Fixup, MCValue Target,
212                            uint64_t &FixedValue);
213
214  void RecordRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout,
215                        const MCFragment *Fragment, const MCFixup &Fixup,
216                        MCValue Target, uint64_t &FixedValue);
217
218  void BindIndirectSymbols(MCAssembler &Asm);
219
220  /// ComputeSymbolTable - Compute the symbol table data
221  ///
222  /// \param StringTable [out] - The string table data.
223  /// \param StringIndexMap [out] - Map from symbol names to offsets in the
224  /// string table.
225  void ComputeSymbolTable(MCAssembler &Asm, SmallString<256> &StringTable,
226                          std::vector<MachSymbolData> &LocalSymbolData,
227                          std::vector<MachSymbolData> &ExternalSymbolData,
228                          std::vector<MachSymbolData> &UndefinedSymbolData);
229
230  void computeSectionAddresses(const MCAssembler &Asm,
231                               const MCAsmLayout &Layout);
232
233  void ExecutePostLayoutBinding(MCAssembler &Asm, const MCAsmLayout &Layout);
234
235  virtual bool IsSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
236                                                      const MCSymbolData &DataA,
237                                                      const MCFragment &FB,
238                                                      bool InSet,
239                                                      bool IsPCRel) const;
240
241  void WriteObject(MCAssembler &Asm, const MCAsmLayout &Layout);
242};
243
244
245/// \brief Construct a new Mach-O writer instance.
246///
247/// This routine takes ownership of the target writer subclass.
248///
249/// \param MOTW - The target specific Mach-O writer subclass.
250/// \param OS - The stream to write to.
251/// \returns The constructed object writer.
252MCObjectWriter *createMachObjectWriter(MCMachObjectTargetWriter *MOTW,
253                                       raw_ostream &OS, bool IsLittleEndian);
254
255} // End llvm namespace
256
257#endif
258