1 //===- llvm/MC/MCMachObjectWriter.h - Mach Object Writer --------*- C++ -*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 
9 #ifndef LLVM_MC_MCMACHOBJECTWRITER_H
10 #define LLVM_MC_MCMACHOBJECTWRITER_H
11 
12 #include "llvm/ADT/DenseMap.h"
13 #include "llvm/ADT/StringRef.h"
14 #include "llvm/BinaryFormat/MachO.h"
15 #include "llvm/MC/MCExpr.h"
16 #include "llvm/MC/MCObjectWriter.h"
17 #include "llvm/MC/MCSection.h"
18 #include "llvm/MC/StringTableBuilder.h"
19 #include "llvm/Support/EndianStream.h"
20 #include <cstdint>
21 #include <memory>
22 #include <string>
23 #include <vector>
24 
25 namespace llvm {
26 
27 class MachObjectWriter;
28 
29 class MCMachObjectTargetWriter : public MCObjectTargetWriter {
30   const unsigned Is64Bit : 1;
31   const uint32_t CPUType;
32 protected:
33   uint32_t CPUSubtype;
34 public:
35   unsigned LocalDifference_RIT;
36 
37 protected:
38   MCMachObjectTargetWriter(bool Is64Bit_, uint32_t CPUType_,
39                            uint32_t CPUSubtype_);
40 
41   void setLocalDifferenceRelocationType(unsigned Type) {
42     LocalDifference_RIT = Type;
43   }
44 
45 public:
46   virtual ~MCMachObjectTargetWriter();
47 
48   Triple::ObjectFormatType getFormat() const override { return Triple::MachO; }
49   static bool classof(const MCObjectTargetWriter *W) {
50     return W->getFormat() == Triple::MachO;
51   }
52 
53   /// \name Lifetime Management
54   /// @{
55 
56   virtual void reset() {}
57 
58   /// @}
59 
60   /// \name Accessors
61   /// @{
62 
63   bool is64Bit() const { return Is64Bit; }
64   uint32_t getCPUType() const { return CPUType; }
65   uint32_t getCPUSubtype() const { return CPUSubtype; }
66   unsigned getLocalDifferenceRelocationType() const {
67     return LocalDifference_RIT;
68   }
69 
70   /// @}
71 
72   /// \name API
73   /// @{
74 
75   virtual void recordRelocation(MachObjectWriter *Writer, MCAssembler &Asm,
76                                 const MCAsmLayout &Layout,
77                                 const MCFragment *Fragment,
78                                 const MCFixup &Fixup, MCValue Target,
79                                 uint64_t &FixedValue) = 0;
80 
81   /// @}
82 };
83 
84 class MachObjectWriter : public MCObjectWriter {
85   /// Helper struct for containing some precomputed information on symbols.
86   struct MachSymbolData {
87     const MCSymbol *Symbol;
88     uint64_t StringIndex;
89     uint8_t SectionIndex;
90 
91     // Support lexicographic sorting.
92     bool operator<(const MachSymbolData &RHS) const;
93   };
94 
95   /// The target specific Mach-O writer instance.
96   std::unique_ptr<MCMachObjectTargetWriter> TargetObjectWriter;
97 
98   /// \name Relocation Data
99   /// @{
100 
101   struct RelAndSymbol {
102     const MCSymbol *Sym;
103     MachO::any_relocation_info MRE;
104     RelAndSymbol(const MCSymbol *Sym, const MachO::any_relocation_info &MRE)
105         : Sym(Sym), MRE(MRE) {}
106   };
107 
108   DenseMap<const MCSection *, std::vector<RelAndSymbol>> Relocations;
109   DenseMap<const MCSection *, unsigned> IndirectSymBase;
110 
111   SectionAddrMap SectionAddress;
112 
113   /// @}
114   /// \name Symbol Table Data
115   /// @{
116 
117   StringTableBuilder StringTable{StringTableBuilder::MachO};
118   std::vector<MachSymbolData> LocalSymbolData;
119   std::vector<MachSymbolData> ExternalSymbolData;
120   std::vector<MachSymbolData> UndefinedSymbolData;
121 
122   /// @}
123 
124   MachSymbolData *findSymbolData(const MCSymbol &Sym);
125 
126   void writeWithPadding(StringRef Str, uint64_t Size);
127 
128 public:
129   MachObjectWriter(std::unique_ptr<MCMachObjectTargetWriter> MOTW,
130                    raw_pwrite_stream &OS, bool IsLittleEndian)
131       : TargetObjectWriter(std::move(MOTW)),
132         W(OS, IsLittleEndian ? support::little : support::big) {}
133 
134   support::endian::Writer W;
135 
136   const MCSymbol &findAliasedSymbol(const MCSymbol &Sym) const;
137 
138   /// \name Lifetime management Methods
139   /// @{
140 
141   void reset() override;
142 
143   /// @}
144 
145   /// \name Utility Methods
146   /// @{
147 
148   bool isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind);
149 
150   SectionAddrMap &getSectionAddressMap() { return SectionAddress; }
151 
152   uint64_t getSectionAddress(const MCSection *Sec) const {
153     return SectionAddress.lookup(Sec);
154   }
155   uint64_t getSymbolAddress(const MCSymbol &S, const MCAsmLayout &Layout) const;
156 
157   uint64_t getFragmentAddress(const MCFragment *Fragment,
158                               const MCAsmLayout &Layout) const;
159 
160   uint64_t getPaddingSize(const MCSection *SD, const MCAsmLayout &Layout) const;
161 
162   bool doesSymbolRequireExternRelocation(const MCSymbol &S);
163 
164   /// @}
165 
166   /// \name Target Writer Proxy Accessors
167   /// @{
168 
169   bool is64Bit() const { return TargetObjectWriter->is64Bit(); }
170   bool isX86_64() const {
171     uint32_t CPUType = TargetObjectWriter->getCPUType();
172     return CPUType == MachO::CPU_TYPE_X86_64;
173   }
174 
175   /// @}
176 
177   void writeHeader(MachO::HeaderFileType Type, unsigned NumLoadCommands,
178                    unsigned LoadCommandsSize, bool SubsectionsViaSymbols);
179 
180   /// Write a segment load command.
181   ///
182   /// \param NumSections The number of sections in this segment.
183   /// \param SectionDataSize The total size of the sections.
184   void writeSegmentLoadCommand(StringRef Name, unsigned NumSections,
185                                uint64_t VMAddr, uint64_t VMSize,
186                                uint64_t SectionDataStartOffset,
187                                uint64_t SectionDataSize, uint32_t MaxProt,
188                                uint32_t InitProt);
189 
190   void writeSection(const MCAsmLayout &Layout, const MCSection &Sec,
191                     uint64_t VMAddr, uint64_t FileOffset, unsigned Flags,
192                     uint64_t RelocationsStart, unsigned NumRelocations);
193 
194   void writeSymtabLoadCommand(uint32_t SymbolOffset, uint32_t NumSymbols,
195                               uint32_t StringTableOffset,
196                               uint32_t StringTableSize);
197 
198   void writeDysymtabLoadCommand(
199       uint32_t FirstLocalSymbol, uint32_t NumLocalSymbols,
200       uint32_t FirstExternalSymbol, uint32_t NumExternalSymbols,
201       uint32_t FirstUndefinedSymbol, uint32_t NumUndefinedSymbols,
202       uint32_t IndirectSymbolOffset, uint32_t NumIndirectSymbols);
203 
204   void writeNlist(MachSymbolData &MSD, const MCAsmLayout &Layout);
205 
206   void writeLinkeditLoadCommand(uint32_t Type, uint32_t DataOffset,
207                                 uint32_t DataSize);
208 
209   void writeLinkerOptionsLoadCommand(const std::vector<std::string> &Options);
210 
211   // FIXME: We really need to improve the relocation validation. Basically, we
212   // want to implement a separate computation which evaluates the relocation
213   // entry as the linker would, and verifies that the resultant fixup value is
214   // exactly what the encoder wanted. This will catch several classes of
215   // problems:
216   //
217   //  - Relocation entry bugs, the two algorithms are unlikely to have the same
218   //    exact bug.
219   //
220   //  - Relaxation issues, where we forget to relax something.
221   //
222   //  - Input errors, where something cannot be correctly encoded. 'as' allows
223   //    these through in many cases.
224 
225   // Add a relocation to be output in the object file. At the time this is
226   // called, the symbol indexes are not know, so if the relocation refers
227   // to a symbol it should be passed as \p RelSymbol so that it can be updated
228   // afterwards. If the relocation doesn't refer to a symbol, nullptr should be
229   // used.
230   void addRelocation(const MCSymbol *RelSymbol, const MCSection *Sec,
231                      MachO::any_relocation_info &MRE) {
232     RelAndSymbol P(RelSymbol, MRE);
233     Relocations[Sec].push_back(P);
234   }
235 
236   void recordScatteredRelocation(const MCAssembler &Asm,
237                                  const MCAsmLayout &Layout,
238                                  const MCFragment *Fragment,
239                                  const MCFixup &Fixup, MCValue Target,
240                                  unsigned Log2Size, uint64_t &FixedValue);
241 
242   void recordTLVPRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout,
243                             const MCFragment *Fragment, const MCFixup &Fixup,
244                             MCValue Target, uint64_t &FixedValue);
245 
246   void recordRelocation(MCAssembler &Asm, const MCAsmLayout &Layout,
247                         const MCFragment *Fragment, const MCFixup &Fixup,
248                         MCValue Target, uint64_t &FixedValue) override;
249 
250   void bindIndirectSymbols(MCAssembler &Asm);
251 
252   /// Compute the symbol table data.
253   void computeSymbolTable(MCAssembler &Asm,
254                           std::vector<MachSymbolData> &LocalSymbolData,
255                           std::vector<MachSymbolData> &ExternalSymbolData,
256                           std::vector<MachSymbolData> &UndefinedSymbolData);
257 
258   void computeSectionAddresses(const MCAssembler &Asm,
259                                const MCAsmLayout &Layout);
260 
261   void executePostLayoutBinding(MCAssembler &Asm,
262                                 const MCAsmLayout &Layout) override;
263 
264   bool isSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
265                                               const MCSymbol &A,
266                                               const MCSymbol &B,
267                                               bool InSet) const override;
268 
269   bool isSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
270                                               const MCSymbol &SymA,
271                                               const MCFragment &FB, bool InSet,
272                                               bool IsPCRel) const override;
273 
274   uint64_t writeObject(MCAssembler &Asm, const MCAsmLayout &Layout) override;
275 };
276 
277 /// Construct a new Mach-O writer instance.
278 ///
279 /// This routine takes ownership of the target writer subclass.
280 ///
281 /// \param MOTW - The target specific Mach-O writer subclass.
282 /// \param OS - The stream to write to.
283 /// \returns The constructed object writer.
284 std::unique_ptr<MCObjectWriter>
285 createMachObjectWriter(std::unique_ptr<MCMachObjectTargetWriter> MOTW,
286                        raw_pwrite_stream &OS, bool IsLittleEndian);
287 
288 } // end namespace llvm
289 
290 #endif // LLVM_MC_MCMACHOBJECTWRITER_H
291