1 //===- MachOReader.cpp ------------------------------------------*- 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 #include "MachOReader.h"
10 #include "Object.h"
11 #include "llvm/BinaryFormat/MachO.h"
12 #include "llvm/Object/MachO.h"
13 #include "llvm/Support/Errc.h"
14 #include <memory>
15 
16 using namespace llvm;
17 using namespace llvm::objcopy;
18 using namespace llvm::objcopy::macho;
19 
20 void MachOReader::readHeader(Object &O) const {
21   O.Header.Magic = MachOObj.getHeader().magic;
22   O.Header.CPUType = MachOObj.getHeader().cputype;
23   O.Header.CPUSubType = MachOObj.getHeader().cpusubtype;
24   O.Header.FileType = MachOObj.getHeader().filetype;
25   O.Header.NCmds = MachOObj.getHeader().ncmds;
26   O.Header.SizeOfCmds = MachOObj.getHeader().sizeofcmds;
27   O.Header.Flags = MachOObj.getHeader().flags;
28 }
29 
30 template <typename SectionType>
31 static Section constructSectionCommon(const SectionType &Sec, uint32_t Index) {
32   StringRef SegName(Sec.segname, strnlen(Sec.segname, sizeof(Sec.segname)));
33   StringRef SectName(Sec.sectname, strnlen(Sec.sectname, sizeof(Sec.sectname)));
34   Section S(SegName, SectName);
35   S.Index = Index;
36   S.Addr = Sec.addr;
37   S.Size = Sec.size;
38   S.OriginalOffset = Sec.offset;
39   S.Align = Sec.align;
40   S.RelOff = Sec.reloff;
41   S.NReloc = Sec.nreloc;
42   S.Flags = Sec.flags;
43   S.Reserved1 = Sec.reserved1;
44   S.Reserved2 = Sec.reserved2;
45   S.Reserved3 = 0;
46   return S;
47 }
48 
49 Section constructSection(const MachO::section &Sec, uint32_t Index) {
50   return constructSectionCommon(Sec, Index);
51 }
52 
53 Section constructSection(const MachO::section_64 &Sec, uint32_t Index) {
54   Section S = constructSectionCommon(Sec, Index);
55   S.Reserved3 = Sec.reserved3;
56   return S;
57 }
58 
59 template <typename SectionType, typename SegmentType>
60 Expected<std::vector<std::unique_ptr<Section>>> static extractSections(
61     const object::MachOObjectFile::LoadCommandInfo &LoadCmd,
62     const object::MachOObjectFile &MachOObj, uint32_t &NextSectionIndex) {
63   std::vector<std::unique_ptr<Section>> Sections;
64   for (auto Curr = reinterpret_cast<const SectionType *>(LoadCmd.Ptr +
65                                                          sizeof(SegmentType)),
66             End = reinterpret_cast<const SectionType *>(LoadCmd.Ptr +
67                                                         LoadCmd.C.cmdsize);
68        Curr < End; ++Curr) {
69     SectionType Sec;
70     memcpy((void *)&Sec, Curr, sizeof(SectionType));
71 
72     if (MachOObj.isLittleEndian() != sys::IsLittleEndianHost)
73       MachO::swapStruct(Sec);
74 
75     Sections.push_back(
76         std::make_unique<Section>(constructSection(Sec, NextSectionIndex)));
77 
78     Section &S = *Sections.back();
79 
80     Expected<object::SectionRef> SecRef =
81         MachOObj.getSection(NextSectionIndex++);
82     if (!SecRef)
83       return SecRef.takeError();
84 
85     Expected<ArrayRef<uint8_t>> Data =
86         MachOObj.getSectionContents(SecRef->getRawDataRefImpl());
87     if (!Data)
88       return Data.takeError();
89 
90     S.Content =
91         StringRef(reinterpret_cast<const char *>(Data->data()), Data->size());
92 
93     const uint32_t CPUType = MachOObj.getHeader().cputype;
94     S.Relocations.reserve(S.NReloc);
95     for (auto RI = MachOObj.section_rel_begin(SecRef->getRawDataRefImpl()),
96               RE = MachOObj.section_rel_end(SecRef->getRawDataRefImpl());
97          RI != RE; ++RI) {
98       RelocationInfo R;
99       R.Symbol = nullptr; // We'll fill this field later.
100       R.Info = MachOObj.getRelocation(RI->getRawDataRefImpl());
101       R.Scattered = MachOObj.isRelocationScattered(R.Info);
102       unsigned Type = MachOObj.getAnyRelocationType(R.Info);
103       // TODO Support CPU_TYPE_ARM.
104       R.IsAddend = !R.Scattered && (CPUType == MachO::CPU_TYPE_ARM64 &&
105                                     Type == MachO::ARM64_RELOC_ADDEND);
106       R.Extern = !R.Scattered && MachOObj.getPlainRelocationExternal(R.Info);
107       S.Relocations.push_back(R);
108     }
109 
110     assert(S.NReloc == S.Relocations.size() &&
111            "Incorrect number of relocations");
112   }
113   return std::move(Sections);
114 }
115 
116 Error MachOReader::readLoadCommands(Object &O) const {
117   // For MachO sections indices start from 1.
118   uint32_t NextSectionIndex = 1;
119   for (auto LoadCmd : MachOObj.load_commands()) {
120     LoadCommand LC;
121     switch (LoadCmd.C.cmd) {
122     case MachO::LC_CODE_SIGNATURE:
123       O.CodeSignatureCommandIndex = O.LoadCommands.size();
124       break;
125     case MachO::LC_SEGMENT:
126       if (Expected<std::vector<std::unique_ptr<Section>>> Sections =
127               extractSections<MachO::section, MachO::segment_command>(
128                   LoadCmd, MachOObj, NextSectionIndex))
129         LC.Sections = std::move(*Sections);
130       else
131         return Sections.takeError();
132       break;
133     case MachO::LC_SEGMENT_64:
134       if (Expected<std::vector<std::unique_ptr<Section>>> Sections =
135               extractSections<MachO::section_64, MachO::segment_command_64>(
136                   LoadCmd, MachOObj, NextSectionIndex))
137         LC.Sections = std::move(*Sections);
138       else
139         return Sections.takeError();
140       break;
141     case MachO::LC_SYMTAB:
142       O.SymTabCommandIndex = O.LoadCommands.size();
143       break;
144     case MachO::LC_DYSYMTAB:
145       O.DySymTabCommandIndex = O.LoadCommands.size();
146       break;
147     case MachO::LC_DYLD_INFO:
148     case MachO::LC_DYLD_INFO_ONLY:
149       O.DyLdInfoCommandIndex = O.LoadCommands.size();
150       break;
151     case MachO::LC_DATA_IN_CODE:
152       O.DataInCodeCommandIndex = O.LoadCommands.size();
153       break;
154     case MachO::LC_LINKER_OPTIMIZATION_HINT:
155       O.LinkerOptimizationHintCommandIndex = O.LoadCommands.size();
156       break;
157     case MachO::LC_FUNCTION_STARTS:
158       O.FunctionStartsCommandIndex = O.LoadCommands.size();
159       break;
160     }
161 #define HANDLE_LOAD_COMMAND(LCName, LCValue, LCStruct)                         \
162   case MachO::LCName:                                                          \
163     memcpy((void *)&(LC.MachOLoadCommand.LCStruct##_data), LoadCmd.Ptr,        \
164            sizeof(MachO::LCStruct));                                           \
165     if (MachOObj.isLittleEndian() != sys::IsLittleEndianHost)                  \
166       MachO::swapStruct(LC.MachOLoadCommand.LCStruct##_data);                  \
167     if (LoadCmd.C.cmdsize > sizeof(MachO::LCStruct))                           \
168       LC.Payload = ArrayRef<uint8_t>(                                          \
169           reinterpret_cast<uint8_t *>(const_cast<char *>(LoadCmd.Ptr)) +       \
170               sizeof(MachO::LCStruct),                                         \
171           LoadCmd.C.cmdsize - sizeof(MachO::LCStruct));                        \
172     break;
173 
174     switch (LoadCmd.C.cmd) {
175     default:
176       memcpy((void *)&(LC.MachOLoadCommand.load_command_data), LoadCmd.Ptr,
177              sizeof(MachO::load_command));
178       if (MachOObj.isLittleEndian() != sys::IsLittleEndianHost)
179         MachO::swapStruct(LC.MachOLoadCommand.load_command_data);
180       if (LoadCmd.C.cmdsize > sizeof(MachO::load_command))
181         LC.Payload = ArrayRef<uint8_t>(
182             reinterpret_cast<uint8_t *>(const_cast<char *>(LoadCmd.Ptr)) +
183                 sizeof(MachO::load_command),
184             LoadCmd.C.cmdsize - sizeof(MachO::load_command));
185       break;
186 #include "llvm/BinaryFormat/MachO.def"
187     }
188     O.LoadCommands.push_back(std::move(LC));
189   }
190   return Error::success();
191 }
192 
193 template <typename nlist_t>
194 SymbolEntry constructSymbolEntry(StringRef StrTable, const nlist_t &nlist) {
195   assert(nlist.n_strx < StrTable.size() &&
196          "n_strx exceeds the size of the string table");
197   SymbolEntry SE;
198   SE.Name = StringRef(StrTable.data() + nlist.n_strx).str();
199   SE.n_type = nlist.n_type;
200   SE.n_sect = nlist.n_sect;
201   SE.n_desc = nlist.n_desc;
202   SE.n_value = nlist.n_value;
203   return SE;
204 }
205 
206 void MachOReader::readSymbolTable(Object &O) const {
207   StringRef StrTable = MachOObj.getStringTableData();
208   for (auto Symbol : MachOObj.symbols()) {
209     SymbolEntry SE =
210         (MachOObj.is64Bit()
211              ? constructSymbolEntry(StrTable, MachOObj.getSymbol64TableEntry(
212                                                   Symbol.getRawDataRefImpl()))
213              : constructSymbolEntry(StrTable, MachOObj.getSymbolTableEntry(
214                                                   Symbol.getRawDataRefImpl())));
215 
216     O.SymTable.Symbols.push_back(std::make_unique<SymbolEntry>(SE));
217   }
218 }
219 
220 void MachOReader::setSymbolInRelocationInfo(Object &O) const {
221   std::vector<const Section *> Sections;
222   for (auto &LC : O.LoadCommands)
223     for (std::unique_ptr<Section> &Sec : LC.Sections)
224       Sections.push_back(Sec.get());
225 
226   for (LoadCommand &LC : O.LoadCommands)
227     for (std::unique_ptr<Section> &Sec : LC.Sections)
228       for (auto &Reloc : Sec->Relocations)
229         if (!Reloc.Scattered && !Reloc.IsAddend) {
230           const uint32_t SymbolNum =
231               Reloc.getPlainRelocationSymbolNum(MachOObj.isLittleEndian());
232           if (Reloc.Extern) {
233             Reloc.Symbol = O.SymTable.getSymbolByIndex(SymbolNum);
234           } else {
235             // FIXME: Refactor error handling in MachOReader and report an error
236             // if we encounter an invalid relocation.
237             assert(SymbolNum >= 1 && SymbolNum <= Sections.size() &&
238                    "Invalid section index.");
239             Reloc.Sec = Sections[SymbolNum - 1];
240           }
241         }
242 }
243 
244 void MachOReader::readRebaseInfo(Object &O) const {
245   O.Rebases.Opcodes = MachOObj.getDyldInfoRebaseOpcodes();
246 }
247 
248 void MachOReader::readBindInfo(Object &O) const {
249   O.Binds.Opcodes = MachOObj.getDyldInfoBindOpcodes();
250 }
251 
252 void MachOReader::readWeakBindInfo(Object &O) const {
253   O.WeakBinds.Opcodes = MachOObj.getDyldInfoWeakBindOpcodes();
254 }
255 
256 void MachOReader::readLazyBindInfo(Object &O) const {
257   O.LazyBinds.Opcodes = MachOObj.getDyldInfoLazyBindOpcodes();
258 }
259 
260 void MachOReader::readExportInfo(Object &O) const {
261   O.Exports.Trie = MachOObj.getDyldInfoExportsTrie();
262 }
263 
264 void MachOReader::readLinkData(Object &O, Optional<size_t> LCIndex,
265                                LinkData &LD) const {
266   if (!LCIndex)
267     return;
268   const MachO::linkedit_data_command &LC =
269       O.LoadCommands[*LCIndex].MachOLoadCommand.linkedit_data_command_data;
270   LD.Data =
271       arrayRefFromStringRef(MachOObj.getData().substr(LC.dataoff, LC.datasize));
272 }
273 
274 void MachOReader::readCodeSignature(Object &O) const {
275   return readLinkData(O, O.CodeSignatureCommandIndex, O.CodeSignature);
276 }
277 
278 void MachOReader::readDataInCodeData(Object &O) const {
279   return readLinkData(O, O.DataInCodeCommandIndex, O.DataInCode);
280 }
281 
282 void MachOReader::readLinkerOptimizationHint(Object &O) const {
283   return readLinkData(O, O.LinkerOptimizationHintCommandIndex,
284                       O.LinkerOptimizationHint);
285 }
286 
287 void MachOReader::readFunctionStartsData(Object &O) const {
288   return readLinkData(O, O.FunctionStartsCommandIndex, O.FunctionStarts);
289 }
290 
291 void MachOReader::readIndirectSymbolTable(Object &O) const {
292   MachO::dysymtab_command DySymTab = MachOObj.getDysymtabLoadCommand();
293   constexpr uint32_t AbsOrLocalMask =
294       MachO::INDIRECT_SYMBOL_LOCAL | MachO::INDIRECT_SYMBOL_ABS;
295   for (uint32_t i = 0; i < DySymTab.nindirectsyms; ++i) {
296     uint32_t Index = MachOObj.getIndirectSymbolTableEntry(DySymTab, i);
297     if ((Index & AbsOrLocalMask) != 0)
298       O.IndirectSymTable.Symbols.emplace_back(Index, None);
299     else
300       O.IndirectSymTable.Symbols.emplace_back(
301           Index, O.SymTable.getSymbolByIndex(Index));
302   }
303 }
304 
305 void MachOReader::readSwiftVersion(Object &O) const {
306   struct ObjCImageInfo {
307     uint32_t Version;
308     uint32_t Flags;
309   } ImageInfo;
310 
311   for (const LoadCommand &LC : O.LoadCommands)
312     for (const std::unique_ptr<Section> &Sec : LC.Sections)
313       if (Sec->Sectname == "__objc_imageinfo" &&
314           (Sec->Segname == "__DATA" || Sec->Segname == "__DATA_CONST" ||
315            Sec->Segname == "__DATA_DIRTY") &&
316           Sec->Content.size() >= sizeof(ObjCImageInfo)) {
317         memcpy(&ImageInfo, Sec->Content.data(), sizeof(ObjCImageInfo));
318         if (MachOObj.isLittleEndian() != sys::IsLittleEndianHost) {
319           sys::swapByteOrder(ImageInfo.Version);
320           sys::swapByteOrder(ImageInfo.Flags);
321         }
322         O.SwiftVersion = (ImageInfo.Flags >> 8) & 0xff;
323         return;
324       }
325 }
326 
327 Expected<std::unique_ptr<Object>> MachOReader::create() const {
328   auto Obj = std::make_unique<Object>();
329   readHeader(*Obj);
330   if (Error E = readLoadCommands(*Obj))
331     return std::move(E);
332   readSymbolTable(*Obj);
333   setSymbolInRelocationInfo(*Obj);
334   readRebaseInfo(*Obj);
335   readBindInfo(*Obj);
336   readWeakBindInfo(*Obj);
337   readLazyBindInfo(*Obj);
338   readExportInfo(*Obj);
339   readCodeSignature(*Obj);
340   readDataInCodeData(*Obj);
341   readLinkerOptimizationHint(*Obj);
342   readFunctionStartsData(*Obj);
343   readIndirectSymbolTable(*Obj);
344   readSwiftVersion(*Obj);
345   return std::move(Obj);
346 }
347