1 //===- MachOYAML.h - Mach-O YAMLIO implementation ---------------*- 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 /// \file
10 /// This file declares classes for handling the YAML representation
11 /// of Mach-O.
12 ///
13 //===----------------------------------------------------------------------===//
14 
15 #ifndef LLVM_OBJECTYAML_MACHOYAML_H
16 #define LLVM_OBJECTYAML_MACHOYAML_H
17 
18 #include "llvm/ADT/StringRef.h"
19 #include "llvm/BinaryFormat/MachO.h"
20 #include "llvm/ObjectYAML/DWARFYAML.h"
21 #include "llvm/ObjectYAML/YAML.h"
22 #include "llvm/Support/YAMLTraits.h"
23 #include <cstdint>
24 #include <optional>
25 #include <string>
26 #include <vector>
27 
28 namespace llvm {
29 namespace MachOYAML {
30 
31 struct Relocation {
32   // Offset in the section to what is being relocated.
33   llvm::yaml::Hex32 address;
34   // Symbol index if r_extern == 1 else section index.
35   uint32_t symbolnum;
36   bool is_pcrel;
37   // Real length = 2 ^ length.
38   uint8_t length;
39   bool is_extern;
40   uint8_t type;
41   bool is_scattered;
42   int32_t value;
43 };
44 
45 struct Section {
46   char sectname[16];
47   char segname[16];
48   llvm::yaml::Hex64 addr;
49   uint64_t size;
50   llvm::yaml::Hex32 offset;
51   uint32_t align;
52   llvm::yaml::Hex32 reloff;
53   uint32_t nreloc;
54   llvm::yaml::Hex32 flags;
55   llvm::yaml::Hex32 reserved1;
56   llvm::yaml::Hex32 reserved2;
57   llvm::yaml::Hex32 reserved3;
58   std::optional<llvm::yaml::BinaryRef> content;
59   std::vector<Relocation> relocations;
60 };
61 
62 struct FileHeader {
63   llvm::yaml::Hex32 magic;
64   llvm::yaml::Hex32 cputype;
65   llvm::yaml::Hex32 cpusubtype;
66   llvm::yaml::Hex32 filetype;
67   uint32_t ncmds;
68   uint32_t sizeofcmds;
69   llvm::yaml::Hex32 flags;
70   llvm::yaml::Hex32 reserved;
71 };
72 
73 struct LoadCommand {
74   virtual ~LoadCommand();
75 
76   llvm::MachO::macho_load_command Data;
77   std::vector<Section> Sections;
78   std::vector<MachO::build_tool_version> Tools;
79   std::vector<llvm::yaml::Hex8> PayloadBytes;
80   std::string Content;
81   uint64_t ZeroPadBytes;
82 };
83 
84 struct NListEntry {
85   uint32_t n_strx;
86   llvm::yaml::Hex8 n_type;
87   uint8_t n_sect;
88   uint16_t n_desc;
89   uint64_t n_value;
90 };
91 
92 struct RebaseOpcode {
93   MachO::RebaseOpcode Opcode;
94   uint8_t Imm;
95   std::vector<yaml::Hex64> ExtraData;
96 };
97 
98 struct BindOpcode {
99   MachO::BindOpcode Opcode;
100   uint8_t Imm;
101   std::vector<yaml::Hex64> ULEBExtraData;
102   std::vector<int64_t> SLEBExtraData;
103   StringRef Symbol;
104 };
105 
106 struct ExportEntry {
107   uint64_t TerminalSize = 0;
108   uint64_t NodeOffset = 0;
109   std::string Name;
110   llvm::yaml::Hex64 Flags = 0;
111   llvm::yaml::Hex64 Address = 0;
112   llvm::yaml::Hex64 Other = 0;
113   std::string ImportName;
114   std::vector<MachOYAML::ExportEntry> Children;
115 };
116 
117 struct DataInCodeEntry {
118   llvm::yaml::Hex32 Offset;
119   uint16_t Length;
120   llvm::yaml::Hex16 Kind;
121 };
122 
123 struct LinkEditData {
124   std::vector<MachOYAML::RebaseOpcode> RebaseOpcodes;
125   std::vector<MachOYAML::BindOpcode> BindOpcodes;
126   std::vector<MachOYAML::BindOpcode> WeakBindOpcodes;
127   std::vector<MachOYAML::BindOpcode> LazyBindOpcodes;
128   MachOYAML::ExportEntry ExportTrie;
129   std::vector<NListEntry> NameList;
130   std::vector<StringRef> StringTable;
131   std::vector<yaml::Hex32> IndirectSymbols;
132   std::vector<yaml::Hex64> FunctionStarts;
133   std::vector<DataInCodeEntry> DataInCode;
134   std::vector<yaml::Hex8> ChainedFixups;
135 
136   bool isEmpty() const;
137 };
138 
139 struct Object {
140   bool IsLittleEndian;
141   FileHeader Header;
142   std::vector<LoadCommand> LoadCommands;
143   std::vector<Section> Sections;
144   LinkEditData LinkEdit;
145   std::optional<llvm::yaml::BinaryRef> RawLinkEditSegment;
146   DWARFYAML::Data DWARF;
147 };
148 
149 struct FatHeader {
150   llvm::yaml::Hex32 magic;
151   uint32_t nfat_arch;
152 };
153 
154 struct FatArch {
155   llvm::yaml::Hex32 cputype;
156   llvm::yaml::Hex32 cpusubtype;
157   llvm::yaml::Hex64 offset;
158   uint64_t size;
159   uint32_t align;
160   llvm::yaml::Hex32 reserved;
161 };
162 
163 struct UniversalBinary {
164   FatHeader Header;
165   std::vector<FatArch> FatArchs;
166   std::vector<Object> Slices;
167 };
168 
169 } // end namespace MachOYAML
170 } // end namespace llvm
171 
172 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachOYAML::LoadCommand)
LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachOYAML::Relocation)173 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachOYAML::Relocation)
174 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachOYAML::Section)
175 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachOYAML::RebaseOpcode)
176 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachOYAML::BindOpcode)
177 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachOYAML::ExportEntry)
178 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachOYAML::NListEntry)
179 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachOYAML::Object)
180 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachOYAML::FatArch)
181 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachOYAML::DataInCodeEntry)
182 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::MachO::build_tool_version)
183 
184 namespace llvm {
185 
186 class raw_ostream;
187 
188 namespace yaml {
189 
190 template <> struct MappingTraits<MachOYAML::FileHeader> {
191   static void mapping(IO &IO, MachOYAML::FileHeader &FileHeader);
192 };
193 
194 template <> struct MappingTraits<MachOYAML::Object> {
195   static void mapping(IO &IO, MachOYAML::Object &Object);
196 };
197 
198 template <> struct MappingTraits<MachOYAML::FatHeader> {
199   static void mapping(IO &IO, MachOYAML::FatHeader &FatHeader);
200 };
201 
202 template <> struct MappingTraits<MachOYAML::FatArch> {
203   static void mapping(IO &IO, MachOYAML::FatArch &FatArch);
204 };
205 
206 template <> struct MappingTraits<MachOYAML::UniversalBinary> {
207   static void mapping(IO &IO, MachOYAML::UniversalBinary &UniversalBinary);
208 };
209 
210 template <> struct MappingTraits<MachOYAML::LoadCommand> {
211   static void mapping(IO &IO, MachOYAML::LoadCommand &LoadCommand);
212 };
213 
214 template <> struct MappingTraits<MachOYAML::LinkEditData> {
215   static void mapping(IO &IO, MachOYAML::LinkEditData &LinkEditData);
216 };
217 
218 template <> struct MappingTraits<MachOYAML::RebaseOpcode> {
219   static void mapping(IO &IO, MachOYAML::RebaseOpcode &RebaseOpcode);
220 };
221 
222 template <> struct MappingTraits<MachOYAML::BindOpcode> {
223   static void mapping(IO &IO, MachOYAML::BindOpcode &BindOpcode);
224 };
225 
226 template <> struct MappingTraits<MachOYAML::ExportEntry> {
227   static void mapping(IO &IO, MachOYAML::ExportEntry &ExportEntry);
228 };
229 
230 template <> struct MappingTraits<MachOYAML::Relocation> {
231   static void mapping(IO &IO, MachOYAML::Relocation &R);
232 };
233 
234 template <> struct MappingTraits<MachOYAML::Section> {
235   static void mapping(IO &IO, MachOYAML::Section &Section);
236   static std::string validate(IO &io, MachOYAML::Section &Section);
237 };
238 
239 template <> struct MappingTraits<MachOYAML::NListEntry> {
240   static void mapping(IO &IO, MachOYAML::NListEntry &NListEntry);
241 };
242 
243 template <> struct MappingTraits<MachO::build_tool_version> {
244   static void mapping(IO &IO, MachO::build_tool_version &tool);
245 };
246 
247 template <> struct MappingTraits<MachOYAML::DataInCodeEntry> {
248   static void mapping(IO &IO, MachOYAML::DataInCodeEntry &DataInCodeEntry);
249 };
250 
251 #define HANDLE_LOAD_COMMAND(LCName, LCValue, LCStruct)                         \
252   io.enumCase(value, #LCName, MachO::LCName);
253 
254 template <> struct ScalarEnumerationTraits<MachO::LoadCommandType> {
255   static void enumeration(IO &io, MachO::LoadCommandType &value) {
256 #include "llvm/BinaryFormat/MachO.def"
257     io.enumFallback<Hex32>(value);
258   }
259 };
260 
261 #define ENUM_CASE(Enum) io.enumCase(value, #Enum, MachO::Enum);
262 
263 template <> struct ScalarEnumerationTraits<MachO::RebaseOpcode> {
264   static void enumeration(IO &io, MachO::RebaseOpcode &value) {
265     ENUM_CASE(REBASE_OPCODE_DONE)
266     ENUM_CASE(REBASE_OPCODE_SET_TYPE_IMM)
267     ENUM_CASE(REBASE_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB)
268     ENUM_CASE(REBASE_OPCODE_ADD_ADDR_ULEB)
269     ENUM_CASE(REBASE_OPCODE_ADD_ADDR_IMM_SCALED)
270     ENUM_CASE(REBASE_OPCODE_DO_REBASE_IMM_TIMES)
271     ENUM_CASE(REBASE_OPCODE_DO_REBASE_ULEB_TIMES)
272     ENUM_CASE(REBASE_OPCODE_DO_REBASE_ADD_ADDR_ULEB)
273     ENUM_CASE(REBASE_OPCODE_DO_REBASE_ULEB_TIMES_SKIPPING_ULEB)
274     io.enumFallback<Hex8>(value);
275   }
276 };
277 
278 template <> struct ScalarEnumerationTraits<MachO::BindOpcode> {
279   static void enumeration(IO &io, MachO::BindOpcode &value) {
280     ENUM_CASE(BIND_OPCODE_DONE)
281     ENUM_CASE(BIND_OPCODE_SET_DYLIB_ORDINAL_IMM)
282     ENUM_CASE(BIND_OPCODE_SET_DYLIB_ORDINAL_ULEB)
283     ENUM_CASE(BIND_OPCODE_SET_DYLIB_SPECIAL_IMM)
284     ENUM_CASE(BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM)
285     ENUM_CASE(BIND_OPCODE_SET_TYPE_IMM)
286     ENUM_CASE(BIND_OPCODE_SET_ADDEND_SLEB)
287     ENUM_CASE(BIND_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB)
288     ENUM_CASE(BIND_OPCODE_ADD_ADDR_ULEB)
289     ENUM_CASE(BIND_OPCODE_DO_BIND)
290     ENUM_CASE(BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB)
291     ENUM_CASE(BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED)
292     ENUM_CASE(BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB)
293     io.enumFallback<Hex8>(value);
294   }
295 };
296 
297 // This trait is used for 16-byte chars in Mach structures used for strings
298 using char_16 = char[16];
299 
300 template <> struct ScalarTraits<char_16> {
301   static void output(const char_16 &Val, void *, raw_ostream &Out);
302   static StringRef input(StringRef Scalar, void *, char_16 &Val);
303   static QuotingType mustQuote(StringRef S);
304 };
305 
306 // This trait is used for UUIDs. It reads and writes them matching otool's
307 // formatting style.
308 using uuid_t = raw_ostream::uuid_t;
309 
310 template <> struct ScalarTraits<uuid_t> {
311   static void output(const uuid_t &Val, void *, raw_ostream &Out);
312   static StringRef input(StringRef Scalar, void *, uuid_t &Val);
313   static QuotingType mustQuote(StringRef S);
314 };
315 
316 // Load Command struct mapping traits
317 
318 #define LOAD_COMMAND_STRUCT(LCStruct)                                          \
319   template <> struct MappingTraits<MachO::LCStruct> {                          \
320     static void mapping(IO &IO, MachO::LCStruct &LoadCommand);                 \
321   };
322 
323 #include "llvm/BinaryFormat/MachO.def"
324 
325 // Extra structures used by load commands
326 template <> struct MappingTraits<MachO::dylib> {
327   static void mapping(IO &IO, MachO::dylib &LoadCommand);
328 };
329 
330 template <> struct MappingTraits<MachO::fvmlib> {
331   static void mapping(IO &IO, MachO::fvmlib &LoadCommand);
332 };
333 
334 template <> struct MappingTraits<MachO::section> {
335   static void mapping(IO &IO, MachO::section &LoadCommand);
336 };
337 
338 template <> struct MappingTraits<MachO::section_64> {
339   static void mapping(IO &IO, MachO::section_64 &LoadCommand);
340 };
341 
342 } // end namespace yaml
343 
344 } // end namespace llvm
345 
346 #endif // LLVM_OBJECTYAML_MACHOYAML_H
347