1 //===- MachO.h - MachO object file implementation ---------------*- 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 file declares the MachOObjectFile class, which implement the ObjectFile
11 // interface for MachO files.
12 //
13 //===----------------------------------------------------------------------===//
14
15 #ifndef LLVM_OBJECT_MACHO_H
16 #define LLVM_OBJECT_MACHO_H
17
18 #include "llvm/ADT/ArrayRef.h"
19 #include "llvm/ADT/SmallVector.h"
20 #include "llvm/ADT/Triple.h"
21 #include "llvm/Object/ObjectFile.h"
22 #include "llvm/Support/MachO.h"
23
24 namespace llvm {
25 namespace object {
26
27 /// DiceRef - This is a value type class that represents a single
28 /// data in code entry in the table in a Mach-O object file.
29 class DiceRef {
30 DataRefImpl DicePimpl;
31 const ObjectFile *OwningObject;
32
33 public:
DiceRef()34 DiceRef() : OwningObject(nullptr) { }
35
36 DiceRef(DataRefImpl DiceP, const ObjectFile *Owner);
37
38 bool operator==(const DiceRef &Other) const;
39 bool operator<(const DiceRef &Other) const;
40
41 void moveNext();
42
43 std::error_code getOffset(uint32_t &Result) const;
44 std::error_code getLength(uint16_t &Result) const;
45 std::error_code getKind(uint16_t &Result) const;
46
47 DataRefImpl getRawDataRefImpl() const;
48 const ObjectFile *getObjectFile() const;
49 };
50 typedef content_iterator<DiceRef> dice_iterator;
51
52 class MachOObjectFile : public ObjectFile {
53 public:
54 struct LoadCommandInfo {
55 const char *Ptr; // Where in memory the load command is.
56 MachO::load_command C; // The command itself.
57 };
58
59 MachOObjectFile(std::unique_ptr<MemoryBuffer> Object, bool IsLittleEndian,
60 bool Is64Bits, std::error_code &EC);
61
62 void moveSymbolNext(DataRefImpl &Symb) const override;
63 std::error_code getSymbolName(DataRefImpl Symb,
64 StringRef &Res) const override;
65
66 // MachO specific.
67 std::error_code getIndirectName(DataRefImpl Symb, StringRef &Res) const;
68
69 std::error_code getSymbolAddress(DataRefImpl Symb,
70 uint64_t &Res) const override;
71 std::error_code getSymbolAlignment(DataRefImpl Symb,
72 uint32_t &Res) const override;
73 std::error_code getSymbolSize(DataRefImpl Symb, uint64_t &Res) const override;
74 std::error_code getSymbolType(DataRefImpl Symb,
75 SymbolRef::Type &Res) const override;
76 uint32_t getSymbolFlags(DataRefImpl Symb) const override;
77 std::error_code getSymbolSection(DataRefImpl Symb,
78 section_iterator &Res) const override;
79
80 void moveSectionNext(DataRefImpl &Sec) const override;
81 std::error_code getSectionName(DataRefImpl Sec,
82 StringRef &Res) const override;
83 std::error_code getSectionAddress(DataRefImpl Sec,
84 uint64_t &Res) const override;
85 std::error_code getSectionSize(DataRefImpl Sec, uint64_t &Res) const override;
86 std::error_code getSectionContents(DataRefImpl Sec,
87 StringRef &Res) const override;
88 std::error_code getSectionAlignment(DataRefImpl Sec,
89 uint64_t &Res) const override;
90 std::error_code isSectionText(DataRefImpl Sec, bool &Res) const override;
91 std::error_code isSectionData(DataRefImpl Sec, bool &Res) const override;
92 std::error_code isSectionBSS(DataRefImpl Sec, bool &Res) const override;
93 std::error_code isSectionRequiredForExecution(DataRefImpl Sec,
94 bool &Res) const override;
95 std::error_code isSectionVirtual(DataRefImpl Sec, bool &Res) const override;
96 std::error_code isSectionZeroInit(DataRefImpl Sec, bool &Res) const override;
97 std::error_code isSectionReadOnlyData(DataRefImpl Sec,
98 bool &Res) const override;
99 std::error_code sectionContainsSymbol(DataRefImpl Sec, DataRefImpl Symb,
100 bool &Result) const override;
101 relocation_iterator section_rel_begin(DataRefImpl Sec) const override;
102 relocation_iterator section_rel_end(DataRefImpl Sec) const override;
103
104 void moveRelocationNext(DataRefImpl &Rel) const override;
105 std::error_code getRelocationAddress(DataRefImpl Rel,
106 uint64_t &Res) const override;
107 std::error_code getRelocationOffset(DataRefImpl Rel,
108 uint64_t &Res) const override;
109 symbol_iterator getRelocationSymbol(DataRefImpl Rel) const override;
110 std::error_code getRelocationType(DataRefImpl Rel,
111 uint64_t &Res) const override;
112 std::error_code
113 getRelocationTypeName(DataRefImpl Rel,
114 SmallVectorImpl<char> &Result) const override;
115 std::error_code
116 getRelocationValueString(DataRefImpl Rel,
117 SmallVectorImpl<char> &Result) const override;
118 std::error_code getRelocationHidden(DataRefImpl Rel,
119 bool &Result) const override;
120
121 std::error_code getLibraryNext(DataRefImpl LibData,
122 LibraryRef &Res) const override;
123 std::error_code getLibraryPath(DataRefImpl LibData,
124 StringRef &Res) const override;
125
126 // MachO specific.
127 std::error_code getLibraryShortNameByIndex(unsigned Index, StringRef &Res);
128
129 // TODO: Would be useful to have an iterator based version
130 // of the load command interface too.
131
132 basic_symbol_iterator symbol_begin_impl() const override;
133 basic_symbol_iterator symbol_end_impl() const override;
134
135 // MachO specific.
136 basic_symbol_iterator getSymbolByIndex(unsigned Index) const;
137
138 section_iterator section_begin() const override;
139 section_iterator section_end() const override;
140
141 library_iterator needed_library_begin() const override;
142 library_iterator needed_library_end() const override;
143
144 uint8_t getBytesInAddress() const override;
145
146 StringRef getFileFormatName() const override;
147 unsigned getArch() const override;
148
149 StringRef getLoadName() const override;
150
151 relocation_iterator section_rel_begin(unsigned Index) const;
152 relocation_iterator section_rel_end(unsigned Index) const;
153
154 dice_iterator begin_dices() const;
155 dice_iterator end_dices() const;
156
157 // In a MachO file, sections have a segment name. This is used in the .o
158 // files. They have a single segment, but this field specifies which segment
159 // a section should be put in in the final object.
160 StringRef getSectionFinalSegmentName(DataRefImpl Sec) const;
161
162 // Names are stored as 16 bytes. These returns the raw 16 bytes without
163 // interpreting them as a C string.
164 ArrayRef<char> getSectionRawName(DataRefImpl Sec) const;
165 ArrayRef<char> getSectionRawFinalSegmentName(DataRefImpl Sec) const;
166
167 // MachO specific Info about relocations.
168 bool isRelocationScattered(const MachO::any_relocation_info &RE) const;
169 unsigned getPlainRelocationSymbolNum(
170 const MachO::any_relocation_info &RE) const;
171 bool getPlainRelocationExternal(const MachO::any_relocation_info &RE) const;
172 bool getScatteredRelocationScattered(
173 const MachO::any_relocation_info &RE) const;
174 uint32_t getScatteredRelocationValue(
175 const MachO::any_relocation_info &RE) const;
176 unsigned getAnyRelocationAddress(const MachO::any_relocation_info &RE) const;
177 unsigned getAnyRelocationPCRel(const MachO::any_relocation_info &RE) const;
178 unsigned getAnyRelocationLength(const MachO::any_relocation_info &RE) const;
179 unsigned getAnyRelocationType(const MachO::any_relocation_info &RE) const;
180 SectionRef getRelocationSection(const MachO::any_relocation_info &RE) const;
181
182 // Walk load commands.
183 LoadCommandInfo getFirstLoadCommandInfo() const;
184 LoadCommandInfo getNextLoadCommandInfo(const LoadCommandInfo &L) const;
185
186 // MachO specific structures.
187 MachO::section getSection(DataRefImpl DRI) const;
188 MachO::section_64 getSection64(DataRefImpl DRI) const;
189 MachO::section getSection(const LoadCommandInfo &L, unsigned Index) const;
190 MachO::section_64 getSection64(const LoadCommandInfo &L,unsigned Index) const;
191 MachO::nlist getSymbolTableEntry(DataRefImpl DRI) const;
192 MachO::nlist_64 getSymbol64TableEntry(DataRefImpl DRI) const;
193
194 MachO::linkedit_data_command
195 getLinkeditDataLoadCommand(const LoadCommandInfo &L) const;
196 MachO::segment_command
197 getSegmentLoadCommand(const LoadCommandInfo &L) const;
198 MachO::segment_command_64
199 getSegment64LoadCommand(const LoadCommandInfo &L) const;
200 MachO::linker_options_command
201 getLinkerOptionsLoadCommand(const LoadCommandInfo &L) const;
202 MachO::version_min_command
203 getVersionMinLoadCommand(const LoadCommandInfo &L) const;
204 MachO::dylib_command
205 getDylibIDLoadCommand(const LoadCommandInfo &L) const;
206
207 MachO::any_relocation_info getRelocation(DataRefImpl Rel) const;
208 MachO::data_in_code_entry getDice(DataRefImpl Rel) const;
209 MachO::mach_header getHeader() const;
210 MachO::mach_header_64 getHeader64() const;
211 uint32_t
212 getIndirectSymbolTableEntry(const MachO::dysymtab_command &DLC,
213 unsigned Index) const;
214 MachO::data_in_code_entry getDataInCodeTableEntry(uint32_t DataOffset,
215 unsigned Index) const;
216 MachO::symtab_command getSymtabLoadCommand() const;
217 MachO::dysymtab_command getDysymtabLoadCommand() const;
218 MachO::linkedit_data_command getDataInCodeLoadCommand() const;
219
220 StringRef getStringTableData() const;
221 bool is64Bit() const;
222 void ReadULEB128s(uint64_t Index, SmallVectorImpl<uint64_t> &Out) const;
223
224 static StringRef guessLibraryShortName(StringRef Name, bool &isFramework,
225 StringRef &Suffix);
226
227 static Triple::ArchType getArch(uint32_t CPUType);
228 static Triple getArch(uint32_t CPUType, uint32_t CPUSubType);
229 static Triple getArch(StringRef ArchFlag);
230 static Triple getHostArch();
231
classof(const Binary * v)232 static bool classof(const Binary *v) {
233 return v->isMachO();
234 }
235
236 const char *getSectionPointer(DataRefImpl Rel) const;
237
238 private:
239 typedef SmallVector<const char *, 1> SectionList;
240 SectionList Sections;
241 typedef SmallVector<const char *, 1> LibraryList;
242 LibraryList Libraries;
243 typedef SmallVector<StringRef, 1> LibraryShortName;
244 LibraryShortName LibrariesShortNames;
245 const char *SymtabLoadCmd;
246 const char *DysymtabLoadCmd;
247 const char *DataInCodeLoadCmd;
248 };
249
250 /// DiceRef
DiceRef(DataRefImpl DiceP,const ObjectFile * Owner)251 inline DiceRef::DiceRef(DataRefImpl DiceP, const ObjectFile *Owner)
252 : DicePimpl(DiceP) , OwningObject(Owner) {}
253
254 inline bool DiceRef::operator==(const DiceRef &Other) const {
255 return DicePimpl == Other.DicePimpl;
256 }
257
258 inline bool DiceRef::operator<(const DiceRef &Other) const {
259 return DicePimpl < Other.DicePimpl;
260 }
261
moveNext()262 inline void DiceRef::moveNext() {
263 const MachO::data_in_code_entry *P =
264 reinterpret_cast<const MachO::data_in_code_entry *>(DicePimpl.p);
265 DicePimpl.p = reinterpret_cast<uintptr_t>(P + 1);
266 }
267
268 // Since a Mach-O data in code reference, a DiceRef, can only be created when
269 // the OwningObject ObjectFile is a MachOObjectFile a static_cast<> is used for
270 // the methods that get the values of the fields of the reference.
271
getOffset(uint32_t & Result)272 inline std::error_code DiceRef::getOffset(uint32_t &Result) const {
273 const MachOObjectFile *MachOOF =
274 static_cast<const MachOObjectFile *>(OwningObject);
275 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
276 Result = Dice.offset;
277 return object_error::success;
278 }
279
getLength(uint16_t & Result)280 inline std::error_code DiceRef::getLength(uint16_t &Result) const {
281 const MachOObjectFile *MachOOF =
282 static_cast<const MachOObjectFile *>(OwningObject);
283 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
284 Result = Dice.length;
285 return object_error::success;
286 }
287
getKind(uint16_t & Result)288 inline std::error_code DiceRef::getKind(uint16_t &Result) const {
289 const MachOObjectFile *MachOOF =
290 static_cast<const MachOObjectFile *>(OwningObject);
291 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
292 Result = Dice.kind;
293 return object_error::success;
294 }
295
getRawDataRefImpl()296 inline DataRefImpl DiceRef::getRawDataRefImpl() const {
297 return DicePimpl;
298 }
299
getObjectFile()300 inline const ObjectFile *DiceRef::getObjectFile() const {
301 return OwningObject;
302 }
303
304 }
305 }
306
307 #endif
308
309