MachOObjectFile: Get symbol functions ready for 64 bit.
[oota-llvm.git] / lib / Object / MachOObjectFile.cpp
1 //===- MachOObjectFile.cpp - Mach-O object file binding ---------*- 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 defines the MachOObjectFile class, which binds the MachOObject
11 // class to the generic ObjectFile wrapper.
12 //
13 //===----------------------------------------------------------------------===//
14
15 #include "llvm/ADT/Triple.h"
16 #include "llvm/Object/MachOFormat.h"
17 #include "llvm/Object/MachOObject.h"
18 #include "llvm/Object/ObjectFile.h"
19 #include "llvm/Support/MemoryBuffer.h"
20 #include "llvm/Support/MachO.h"
21
22 #include <cctype>
23 #include <cstring>
24 #include <limits>
25
26 using namespace llvm;
27 using namespace object;
28
29 namespace llvm {
30
31 typedef MachOObject::LoadCommandInfo LoadCommandInfo;
32
33 class MachOObjectFile : public ObjectFile {
34 public:
35   MachOObjectFile(MemoryBuffer *Object, MachOObject *MOO, error_code &ec)
36     : ObjectFile(Binary::isMachO, Object, ec),
37       MachOObj(MOO),
38       RegisteredStringTable(std::numeric_limits<uint32_t>::max()) {}
39
40   virtual symbol_iterator begin_symbols() const;
41   virtual symbol_iterator end_symbols() const;
42   virtual section_iterator begin_sections() const;
43   virtual section_iterator end_sections() const;
44
45   virtual uint8_t getBytesInAddress() const;
46   virtual StringRef getFileFormatName() const;
47   virtual unsigned getArch() const;
48
49 protected:
50   virtual error_code getSymbolNext(DataRefImpl Symb, SymbolRef &Res) const;
51   virtual error_code getSymbolName(DataRefImpl Symb, StringRef &Res) const;
52   virtual error_code getSymbolAddress(DataRefImpl Symb, uint64_t &Res) const;
53   virtual error_code getSymbolSize(DataRefImpl Symb, uint64_t &Res) const;
54   virtual error_code getSymbolNMTypeChar(DataRefImpl Symb, char &Res) const;
55   virtual error_code isSymbolInternal(DataRefImpl Symb, bool &Res) const;
56
57   virtual error_code getSectionNext(DataRefImpl Sec, SectionRef &Res) const;
58   virtual error_code getSectionName(DataRefImpl Sec, StringRef &Res) const;
59   virtual error_code getSectionAddress(DataRefImpl Sec, uint64_t &Res) const;
60   virtual error_code getSectionSize(DataRefImpl Sec, uint64_t &Res) const;
61   virtual error_code getSectionContents(DataRefImpl Sec, StringRef &Res) const;
62   virtual error_code isSectionText(DataRefImpl Sec, bool &Res) const;
63
64 private:
65   MachOObject *MachOObj;
66   mutable uint32_t RegisteredStringTable;
67
68   void moveToNextSection(DataRefImpl &DRI) const;
69   void getSymbolTableEntry(DataRefImpl DRI,
70                            InMemoryStruct<macho::SymbolTableEntry> &Res) const;
71   void getSymbol64TableEntry(DataRefImpl DRI,
72                           InMemoryStruct<macho::Symbol64TableEntry> &Res) const;
73   void moveToNextSymbol(DataRefImpl &DRI) const;
74   void getSection(DataRefImpl DRI, InMemoryStruct<macho::Section> &Res) const;
75   void getSection64(DataRefImpl DRI,
76                     InMemoryStruct<macho::Section64> &Res) const;
77 };
78
79 ObjectFile *ObjectFile::createMachOObjectFile(MemoryBuffer *Buffer) {
80   error_code ec;
81   std::string Err;
82   MachOObject *MachOObj = MachOObject::LoadFromBuffer(Buffer, &Err);
83   if (!MachOObj)
84     return NULL;
85   return new MachOObjectFile(Buffer, MachOObj, ec);
86 }
87
88 /*===-- Symbols -----------------------------------------------------------===*/
89
90 void MachOObjectFile::moveToNextSymbol(DataRefImpl &DRI) const {
91   uint32_t LoadCommandCount = MachOObj->getHeader().NumLoadCommands;
92   while (DRI.d.a < LoadCommandCount) {
93     LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a);
94     if (LCI.Command.Type == macho::LCT_Symtab) {
95       InMemoryStruct<macho::SymtabLoadCommand> SymtabLoadCmd;
96       MachOObj->ReadSymtabLoadCommand(LCI, SymtabLoadCmd);
97       if (DRI.d.b < SymtabLoadCmd->NumSymbolTableEntries)
98         return;
99     }
100
101     DRI.d.a++;
102     DRI.d.b = 0;
103   }
104 }
105
106 void MachOObjectFile::getSymbolTableEntry(DataRefImpl DRI,
107     InMemoryStruct<macho::SymbolTableEntry> &Res) const {
108   InMemoryStruct<macho::SymtabLoadCommand> SymtabLoadCmd;
109   LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a);
110   MachOObj->ReadSymtabLoadCommand(LCI, SymtabLoadCmd);
111
112   if (RegisteredStringTable != DRI.d.a) {
113     MachOObj->RegisterStringTable(*SymtabLoadCmd);
114     RegisteredStringTable = DRI.d.a;
115   }
116
117   MachOObj->ReadSymbolTableEntry(SymtabLoadCmd->SymbolTableOffset, DRI.d.b,
118                                  Res);
119 }
120
121 void MachOObjectFile::getSymbol64TableEntry(DataRefImpl DRI,
122     InMemoryStruct<macho::Symbol64TableEntry> &Res) const {
123   InMemoryStruct<macho::SymtabLoadCommand> SymtabLoadCmd;
124   LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a);
125   MachOObj->ReadSymtabLoadCommand(LCI, SymtabLoadCmd);
126
127   if (RegisteredStringTable != DRI.d.a) {
128     MachOObj->RegisterStringTable(*SymtabLoadCmd);
129     RegisteredStringTable = DRI.d.a;
130   }
131
132   MachOObj->ReadSymbol64TableEntry(SymtabLoadCmd->SymbolTableOffset, DRI.d.b,
133                                    Res);
134 }
135
136
137 error_code MachOObjectFile::getSymbolNext(DataRefImpl DRI,
138                                           SymbolRef &Result) const {
139   DRI.d.b++;
140   moveToNextSymbol(DRI);
141   Result = SymbolRef(DRI, this);
142   return object_error::success;
143 }
144
145 error_code MachOObjectFile::getSymbolName(DataRefImpl DRI,
146                                           StringRef &Result) const {
147   if (MachOObj->is64Bit()) {
148     InMemoryStruct<macho::Symbol64TableEntry> Entry;
149     getSymbol64TableEntry(DRI, Entry);
150     Result = MachOObj->getStringAtIndex(Entry->StringIndex);
151   } else {
152     InMemoryStruct<macho::SymbolTableEntry> Entry;
153     getSymbolTableEntry(DRI, Entry);
154     Result = MachOObj->getStringAtIndex(Entry->StringIndex);
155   }
156   return object_error::success;
157 }
158
159 error_code MachOObjectFile::getSymbolAddress(DataRefImpl DRI,
160                                              uint64_t &Result) const {
161   if (MachOObj->is64Bit()) {
162     InMemoryStruct<macho::Symbol64TableEntry> Entry;
163     getSymbol64TableEntry(DRI, Entry);
164     Result = Entry->Value;
165   } else {
166     InMemoryStruct<macho::SymbolTableEntry> Entry;
167     getSymbolTableEntry(DRI, Entry);
168     Result = Entry->Value;
169   }
170   return object_error::success;
171 }
172
173 error_code MachOObjectFile::getSymbolSize(DataRefImpl DRI,
174                                           uint64_t &Result) const {
175   Result = UnknownAddressOrSize;
176   return object_error::success;
177 }
178
179 error_code MachOObjectFile::getSymbolNMTypeChar(DataRefImpl DRI,
180                                                 char &Result) const {
181   uint8_t Type, Flags;
182   if (MachOObj->is64Bit()) {
183     InMemoryStruct<macho::Symbol64TableEntry> Entry;
184     getSymbol64TableEntry(DRI, Entry);
185     Type = Entry->Type;
186     Flags = Entry->Flags;
187   } else {
188     InMemoryStruct<macho::SymbolTableEntry> Entry;
189     getSymbolTableEntry(DRI, Entry);
190     Type = Entry->Type;
191     Flags = Entry->Flags;
192   }
193
194   char Char;
195   switch (Type & macho::STF_TypeMask) {
196     case macho::STT_Undefined:
197       Char = 'u';
198       break;
199     case macho::STT_Absolute:
200     case macho::STT_Section:
201       Char = 's';
202       break;
203     default:
204       Char = '?';
205       break;
206   }
207
208   if (Flags & (macho::STF_External | macho::STF_PrivateExtern))
209     Char = toupper(Char);
210   Result = Char;
211   return object_error::success;
212 }
213
214 error_code MachOObjectFile::isSymbolInternal(DataRefImpl DRI,
215                                              bool &Result) const {
216   if (MachOObj->is64Bit()) {
217     InMemoryStruct<macho::Symbol64TableEntry> Entry;
218     getSymbol64TableEntry(DRI, Entry);
219     Result = Entry->Flags & macho::STF_StabsEntryMask;
220   } else {
221     InMemoryStruct<macho::SymbolTableEntry> Entry;
222     getSymbolTableEntry(DRI, Entry);
223     Result = Entry->Flags & macho::STF_StabsEntryMask;
224   }
225   return object_error::success;
226 }
227
228 ObjectFile::symbol_iterator MachOObjectFile::begin_symbols() const {
229   // DRI.d.a = segment number; DRI.d.b = symbol index.
230   DataRefImpl DRI;
231   DRI.d.a = DRI.d.b = 0;
232   moveToNextSymbol(DRI);
233   return symbol_iterator(SymbolRef(DRI, this));
234 }
235
236 ObjectFile::symbol_iterator MachOObjectFile::end_symbols() const {
237   DataRefImpl DRI;
238   DRI.d.a = MachOObj->getHeader().NumLoadCommands;
239   DRI.d.b = 0;
240   return symbol_iterator(SymbolRef(DRI, this));
241 }
242
243
244 /*===-- Sections ----------------------------------------------------------===*/
245
246 void MachOObjectFile::moveToNextSection(DataRefImpl &DRI) const {
247   uint32_t LoadCommandCount = MachOObj->getHeader().NumLoadCommands;
248   while (DRI.d.a < LoadCommandCount) {
249     LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a);
250     if (LCI.Command.Type == macho::LCT_Segment) {
251       InMemoryStruct<macho::SegmentLoadCommand> SegmentLoadCmd;
252       MachOObj->ReadSegmentLoadCommand(LCI, SegmentLoadCmd);
253       if (DRI.d.b < SegmentLoadCmd->NumSections)
254         return;
255     } else if (LCI.Command.Type == macho::LCT_Segment64) {
256       InMemoryStruct<macho::Segment64LoadCommand> Segment64LoadCmd;
257       MachOObj->ReadSegment64LoadCommand(LCI, Segment64LoadCmd);
258       if (DRI.d.b < Segment64LoadCmd->NumSections)
259         return;
260     }
261
262     DRI.d.a++;
263     DRI.d.b = 0;
264   }
265 }
266
267 error_code MachOObjectFile::getSectionNext(DataRefImpl DRI,
268                                            SectionRef &Result) const {
269   DRI.d.b++;
270   moveToNextSection(DRI);
271   Result = SectionRef(DRI, this);
272   return object_error::success;
273 }
274
275 void
276 MachOObjectFile::getSection(DataRefImpl DRI,
277                             InMemoryStruct<macho::Section> &Res) const {
278   InMemoryStruct<macho::SegmentLoadCommand> SLC;
279   LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a);
280   MachOObj->ReadSegmentLoadCommand(LCI, SLC);
281   MachOObj->ReadSection(LCI, DRI.d.b, Res);
282 }
283
284 void
285 MachOObjectFile::getSection64(DataRefImpl DRI,
286                             InMemoryStruct<macho::Section64> &Res) const {
287   InMemoryStruct<macho::Segment64LoadCommand> SLC;
288   LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a);
289   MachOObj->ReadSegment64LoadCommand(LCI, SLC);
290   MachOObj->ReadSection64(LCI, DRI.d.b, Res);
291 }
292
293 static bool is64BitLoadCommand(const MachOObject *MachOObj, DataRefImpl DRI) {
294   LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a);
295   if (LCI.Command.Type == macho::LCT_Segment64)
296     return true;
297   assert(LCI.Command.Type == macho::LCT_Segment && "Unexpected Type.");
298   return false;
299 }
300
301 error_code MachOObjectFile::getSectionName(DataRefImpl DRI,
302                                            StringRef &Result) const {
303   // FIXME: thread safety.
304   static char result[34];
305   if (is64BitLoadCommand(MachOObj, DRI)) {
306     InMemoryStruct<macho::Segment64LoadCommand> SLC;
307     LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a);
308     MachOObj->ReadSegment64LoadCommand(LCI, SLC);
309     InMemoryStruct<macho::Section64> Sect;
310     MachOObj->ReadSection64(LCI, DRI.d.b, Sect);
311
312     strcpy(result, Sect->SegmentName);
313     strcat(result, ",");
314     strcat(result, Sect->Name);
315   } else {
316     InMemoryStruct<macho::SegmentLoadCommand> SLC;
317     LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a);
318     MachOObj->ReadSegmentLoadCommand(LCI, SLC);
319     InMemoryStruct<macho::Section> Sect;
320     MachOObj->ReadSection(LCI, DRI.d.b, Sect);
321
322     strcpy(result, Sect->SegmentName);
323     strcat(result, ",");
324     strcat(result, Sect->Name);
325   }
326   Result = StringRef(result);
327   return object_error::success;
328 }
329
330 error_code MachOObjectFile::getSectionAddress(DataRefImpl DRI,
331                                               uint64_t &Result) const {
332   if (is64BitLoadCommand(MachOObj, DRI)) {
333     InMemoryStruct<macho::Section64> Sect;
334     getSection64(DRI, Sect);
335     Result = Sect->Address;
336   } else {
337     InMemoryStruct<macho::Section> Sect;
338     getSection(DRI, Sect);
339     Result = Sect->Address;
340   }
341   return object_error::success;
342 }
343
344 error_code MachOObjectFile::getSectionSize(DataRefImpl DRI,
345                                            uint64_t &Result) const {
346   if (is64BitLoadCommand(MachOObj, DRI)) {
347     InMemoryStruct<macho::Section64> Sect;
348     getSection64(DRI, Sect);
349     Result = Sect->Size;
350   } else {
351     InMemoryStruct<macho::Section> Sect;
352     getSection(DRI, Sect);
353     Result = Sect->Size;
354   }
355   return object_error::success;
356 }
357
358 error_code MachOObjectFile::getSectionContents(DataRefImpl DRI,
359                                                StringRef &Result) const {
360   if (is64BitLoadCommand(MachOObj, DRI)) {
361     InMemoryStruct<macho::Section64> Sect;
362     getSection64(DRI, Sect);
363     Result = MachOObj->getData(Sect->Offset, Sect->Size);
364   } else {
365     InMemoryStruct<macho::Section> Sect;
366     getSection(DRI, Sect);
367     Result = MachOObj->getData(Sect->Offset, Sect->Size);
368   }
369   return object_error::success;
370 }
371
372 error_code MachOObjectFile::isSectionText(DataRefImpl DRI,
373                                           bool &Result) const {
374   if (is64BitLoadCommand(MachOObj, DRI)) {
375     InMemoryStruct<macho::Section64> Sect;
376     getSection64(DRI, Sect);
377     Result = !strcmp(Sect->Name, "__text");
378   } else {
379     InMemoryStruct<macho::Section> Sect;
380     getSection(DRI, Sect);
381     Result = !strcmp(Sect->Name, "__text");
382   }
383   return object_error::success;
384 }
385
386 ObjectFile::section_iterator MachOObjectFile::begin_sections() const {
387   DataRefImpl DRI;
388   DRI.d.a = DRI.d.b = 0;
389   moveToNextSection(DRI);
390   return section_iterator(SectionRef(DRI, this));
391 }
392
393 ObjectFile::section_iterator MachOObjectFile::end_sections() const {
394   DataRefImpl DRI;
395   DRI.d.a = MachOObj->getHeader().NumLoadCommands;
396   DRI.d.b = 0;
397   return section_iterator(SectionRef(DRI, this));
398 }
399
400 /*===-- Miscellaneous -----------------------------------------------------===*/
401
402 uint8_t MachOObjectFile::getBytesInAddress() const {
403   return MachOObj->is64Bit() ? 8 : 4;
404 }
405
406 StringRef MachOObjectFile::getFileFormatName() const {
407   if (!MachOObj->is64Bit()) {
408     switch (MachOObj->getHeader().CPUType) {
409     case llvm::MachO::CPUTypeI386:
410       return "Mach-O 32-bit i386";
411     case llvm::MachO::CPUTypeARM:
412       return "Mach-O arm";
413     case llvm::MachO::CPUTypePowerPC:
414       return "Mach-O 32-bit ppc";
415     default:
416       assert((MachOObj->getHeader().CPUType & llvm::MachO::CPUArchABI64) == 0 &&
417              "64-bit object file when we're not 64-bit?");
418       return "Mach-O 32-bit unknown";
419     }
420   }
421
422   switch (MachOObj->getHeader().CPUType) {
423   case llvm::MachO::CPUTypeX86_64:
424     return "Mach-O 64-bit x86-64";
425   case llvm::MachO::CPUTypePowerPC64:
426     return "Mach-O 64-bit ppc64";
427   default:
428     assert((MachOObj->getHeader().CPUType & llvm::MachO::CPUArchABI64) == 1 &&
429            "32-bit object file when we're 64-bit?");
430     return "Mach-O 64-bit unknown";
431   }
432 }
433
434 unsigned MachOObjectFile::getArch() const {
435   switch (MachOObj->getHeader().CPUType) {
436   case llvm::MachO::CPUTypeI386:
437     return Triple::x86;
438   case llvm::MachO::CPUTypeX86_64:
439     return Triple::x86_64;
440   case llvm::MachO::CPUTypeARM:
441     return Triple::arm;
442   case llvm::MachO::CPUTypePowerPC:
443     return Triple::ppc;
444   case llvm::MachO::CPUTypePowerPC64:
445     return Triple::ppc64;
446   default:
447     return Triple::UnknownArch;
448   }
449 }
450
451 } // end namespace llvm
452