xref: /llvm-project/llvm/lib/Object/COFFObjectFile.cpp (revision 206d8a7f480a9ce3e38f6e611303bdc3c0ee2c44) 
1 //===- COFFObjectFile.cpp - COFF 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 COFFObjectFile class.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "llvm/Object/COFF.h"
15 #include "llvm/ADT/StringSwitch.h"
16 #include "llvm/ADT/Triple.h"
17 
18 using namespace llvm;
19 using namespace object;
20 
21 namespace {
22 using support::ulittle8_t;
23 using support::ulittle16_t;
24 using support::ulittle32_t;
25 using support::little16_t;
26 }
27 
28 namespace {
29 // Returns false if size is greater than the buffer size. And sets ec.
30 bool checkSize(const MemoryBuffer *m, error_code &ec, uint64_t size) {
31   if (m->getBufferSize() < size) {
32     ec = object_error::unexpected_eof;
33     return false;
34   }
35   return true;
36 }
37 
38 // Returns false if any bytes in [addr, addr + size) fall outsize of m.
39 bool checkAddr(const MemoryBuffer *m,
40                error_code &ec,
41                uintptr_t addr,
42                uint64_t size) {
43   if (addr + size < addr ||
44       addr + size < size ||
45       addr + size > uintptr_t(m->getBufferEnd())) {
46     ec = object_error::unexpected_eof;
47     return false;
48   }
49   return true;
50 }
51 }
52 
53 const coff_symbol *COFFObjectFile::toSymb(DataRefImpl Symb) const {
54   const coff_symbol *addr = reinterpret_cast<const coff_symbol*>(Symb.p);
55 
56 # ifndef NDEBUG
57   // Verify that the symbol points to a valid entry in the symbol table.
58   uintptr_t offset = uintptr_t(addr) - uintptr_t(base());
59   if (offset < Header->PointerToSymbolTable
60       || offset >= Header->PointerToSymbolTable
61          + (Header->NumberOfSymbols * sizeof(coff_symbol)))
62     report_fatal_error("Symbol was outside of symbol table.");
63 
64   assert((offset - Header->PointerToSymbolTable) % sizeof(coff_symbol)
65          == 0 && "Symbol did not point to the beginning of a symbol");
66 # endif
67 
68   return addr;
69 }
70 
71 const coff_section *COFFObjectFile::toSec(DataRefImpl Sec) const {
72   const coff_section *addr = reinterpret_cast<const coff_section*>(Sec.p);
73 
74 # ifndef NDEBUG
75   // Verify that the section points to a valid entry in the section table.
76   if (addr < SectionTable
77       || addr >= (SectionTable + Header->NumberOfSections))
78     report_fatal_error("Section was outside of section table.");
79 
80   uintptr_t offset = uintptr_t(addr) - uintptr_t(SectionTable);
81   assert(offset % sizeof(coff_section) == 0 &&
82          "Section did not point to the beginning of a section");
83 # endif
84 
85   return addr;
86 }
87 
88 error_code COFFObjectFile::getSymbolNext(DataRefImpl Symb,
89                                          SymbolRef &Result) const {
90   const coff_symbol *symb = toSymb(Symb);
91   symb += 1 + symb->NumberOfAuxSymbols;
92   Symb.p = reinterpret_cast<uintptr_t>(symb);
93   Result = SymbolRef(Symb, this);
94   return object_error::success;
95 }
96 
97  error_code COFFObjectFile::getSymbolName(DataRefImpl Symb,
98                                           StringRef &Result) const {
99   const coff_symbol *symb = toSymb(Symb);
100   // Check for string table entry. First 4 bytes are 0.
101   if (symb->Name.Offset.Zeroes == 0) {
102     uint32_t Offset = symb->Name.Offset.Offset;
103     if (error_code ec = getString(Offset, Result))
104       return ec;
105     return object_error::success;
106   }
107 
108   if (symb->Name.ShortName[7] == 0)
109     // Null terminated, let ::strlen figure out the length.
110     Result = StringRef(symb->Name.ShortName);
111   else
112     // Not null terminated, use all 8 bytes.
113     Result = StringRef(symb->Name.ShortName, 8);
114   return object_error::success;
115 }
116 
117 error_code COFFObjectFile::getSymbolOffset(DataRefImpl Symb,
118                                             uint64_t &Result) const {
119   const coff_symbol *symb = toSymb(Symb);
120   const coff_section *Section = NULL;
121   if (error_code ec = getSection(symb->SectionNumber, Section))
122     return ec;
123   char Type;
124   if (error_code ec = getSymbolNMTypeChar(Symb, Type))
125     return ec;
126   if (Type == 'U' || Type == 'w')
127     Result = UnknownAddressOrSize;
128   else if (Section)
129     Result = Section->VirtualAddress + symb->Value;
130   else
131     Result = symb->Value;
132   return object_error::success;
133 }
134 
135 error_code COFFObjectFile::getSymbolAddress(DataRefImpl Symb,
136                                             uint64_t &Result) const {
137   const coff_symbol *symb = toSymb(Symb);
138   const coff_section *Section = NULL;
139   if (error_code ec = getSection(symb->SectionNumber, Section))
140     return ec;
141   char Type;
142   if (error_code ec = getSymbolNMTypeChar(Symb, Type))
143     return ec;
144   if (Type == 'U' || Type == 'w')
145     Result = UnknownAddressOrSize;
146   else if (Section)
147     Result = reinterpret_cast<uintptr_t>(base() +
148                                          Section->PointerToRawData +
149                                          symb->Value);
150   else
151     Result = reinterpret_cast<uintptr_t>(base() + symb->Value);
152   return object_error::success;
153 }
154 
155 error_code COFFObjectFile::getSymbolType(DataRefImpl Symb,
156                                          SymbolRef::SymbolType &Result) const {
157   const coff_symbol *symb = toSymb(Symb);
158   Result = SymbolRef::ST_Other;
159   if (symb->StorageClass == COFF::IMAGE_SYM_CLASS_EXTERNAL &&
160       symb->SectionNumber == COFF::IMAGE_SYM_UNDEFINED) {
161     Result = SymbolRef::ST_External;
162   } else {
163     if (symb->Type.ComplexType == COFF::IMAGE_SYM_DTYPE_FUNCTION) {
164       Result = SymbolRef::ST_Function;
165     } else {
166       char Type;
167       if (error_code ec = getSymbolNMTypeChar(Symb, Type))
168         return ec;
169       if (Type == 'r' || Type == 'R') {
170         Result = SymbolRef::ST_Data;
171       }
172     }
173   }
174   return object_error::success;
175 }
176 
177 error_code COFFObjectFile::isSymbolGlobal(DataRefImpl Symb,
178                                           bool &Result) const {
179   const coff_symbol *symb = toSymb(Symb);
180   Result = (symb->StorageClass == COFF::IMAGE_SYM_CLASS_EXTERNAL);
181   return object_error::success;
182 }
183 
184 error_code COFFObjectFile::getSymbolSize(DataRefImpl Symb,
185                                          uint64_t &Result) const {
186   // FIXME: Return the correct size. This requires looking at all the symbols
187   //        in the same section as this symbol, and looking for either the next
188   //        symbol, or the end of the section.
189   const coff_symbol *symb = toSymb(Symb);
190   const coff_section *Section = NULL;
191   if (error_code ec = getSection(symb->SectionNumber, Section))
192     return ec;
193   char Type;
194   if (error_code ec = getSymbolNMTypeChar(Symb, Type))
195     return ec;
196   if (Type == 'U' || Type == 'w')
197     Result = UnknownAddressOrSize;
198   else if (Section)
199     Result = Section->SizeOfRawData - symb->Value;
200   else
201     Result = 0;
202   return object_error::success;
203 }
204 
205 error_code COFFObjectFile::getSymbolNMTypeChar(DataRefImpl Symb,
206                                                char &Result) const {
207   const coff_symbol *symb = toSymb(Symb);
208   StringRef name;
209   if (error_code ec = getSymbolName(Symb, name))
210     return ec;
211   char ret = StringSwitch<char>(name)
212     .StartsWith(".debug", 'N')
213     .StartsWith(".sxdata", 'N')
214     .Default('?');
215 
216   if (ret != '?') {
217     Result = ret;
218     return object_error::success;
219   }
220 
221   uint32_t Characteristics = 0;
222   if (symb->SectionNumber > 0) {
223     const coff_section *Section = NULL;
224     if (error_code ec = getSection(symb->SectionNumber, Section))
225       return ec;
226     Characteristics = Section->Characteristics;
227   }
228 
229   switch (symb->SectionNumber) {
230   case COFF::IMAGE_SYM_UNDEFINED:
231     // Check storage classes.
232     if (symb->StorageClass == COFF::IMAGE_SYM_CLASS_WEAK_EXTERNAL) {
233       Result = 'w';
234       return object_error::success; // Don't do ::toupper.
235     } else
236       ret = 'u';
237     break;
238   case COFF::IMAGE_SYM_ABSOLUTE:
239     ret = 'a';
240     break;
241   case COFF::IMAGE_SYM_DEBUG:
242     ret = 'n';
243     break;
244   default:
245     // Check section type.
246     if (Characteristics & COFF::IMAGE_SCN_CNT_CODE)
247       ret = 't';
248     else if (  Characteristics & COFF::IMAGE_SCN_MEM_READ
249             && ~Characteristics & COFF::IMAGE_SCN_MEM_WRITE) // Read only.
250       ret = 'r';
251     else if (Characteristics & COFF::IMAGE_SCN_CNT_INITIALIZED_DATA)
252       ret = 'd';
253     else if (Characteristics & COFF::IMAGE_SCN_CNT_UNINITIALIZED_DATA)
254       ret = 'b';
255     else if (Characteristics & COFF::IMAGE_SCN_LNK_INFO)
256       ret = 'i';
257 
258     // Check for section symbol.
259     else if (  symb->StorageClass == COFF::IMAGE_SYM_CLASS_STATIC
260             && symb->Value == 0)
261        ret = 's';
262   }
263 
264   if (symb->StorageClass == COFF::IMAGE_SYM_CLASS_EXTERNAL)
265     ret = ::toupper(ret);
266 
267   Result = ret;
268   return object_error::success;
269 }
270 
271 error_code COFFObjectFile::isSymbolInternal(DataRefImpl Symb,
272                                             bool &Result) const {
273   Result = false;
274   return object_error::success;
275 }
276 
277 error_code COFFObjectFile::getSectionNext(DataRefImpl Sec,
278                                           SectionRef &Result) const {
279   const coff_section *sec = toSec(Sec);
280   sec += 1;
281   Sec.p = reinterpret_cast<uintptr_t>(sec);
282   Result = SectionRef(Sec, this);
283   return object_error::success;
284 }
285 
286 error_code COFFObjectFile::getSectionName(DataRefImpl Sec,
287                                           StringRef &Result) const {
288   const coff_section *sec = toSec(Sec);
289   StringRef name;
290   if (sec->Name[7] == 0)
291     // Null terminated, let ::strlen figure out the length.
292     name = sec->Name;
293   else
294     // Not null terminated, use all 8 bytes.
295     name = StringRef(sec->Name, 8);
296 
297   // Check for string table entry. First byte is '/'.
298   if (name[0] == '/') {
299     uint32_t Offset;
300     name.substr(1).getAsInteger(10, Offset);
301     if (error_code ec = getString(Offset, name))
302       return ec;
303   }
304 
305   Result = name;
306   return object_error::success;
307 }
308 
309 error_code COFFObjectFile::getSectionAddress(DataRefImpl Sec,
310                                              uint64_t &Result) const {
311   const coff_section *sec = toSec(Sec);
312   Result = sec->VirtualAddress;
313   return object_error::success;
314 }
315 
316 error_code COFFObjectFile::getSectionSize(DataRefImpl Sec,
317                                           uint64_t &Result) const {
318   const coff_section *sec = toSec(Sec);
319   Result = sec->SizeOfRawData;
320   return object_error::success;
321 }
322 
323 error_code COFFObjectFile::getSectionContents(DataRefImpl Sec,
324                                               StringRef &Result) const {
325   const coff_section *sec = toSec(Sec);
326   // The only thing that we need to verify is that the contents is contained
327   // within the file bounds. We don't need to make sure it doesn't cover other
328   // data, as there's nothing that says that is not allowed.
329   uintptr_t con_start = uintptr_t(base()) + sec->PointerToRawData;
330   uintptr_t con_end = con_start + sec->SizeOfRawData;
331   if (con_end >= uintptr_t(Data->getBufferEnd()))
332     return object_error::parse_failed;
333   Result = StringRef(reinterpret_cast<const char*>(con_start),
334                      sec->SizeOfRawData);
335   return object_error::success;
336 }
337 
338 error_code COFFObjectFile::isSectionText(DataRefImpl Sec,
339                                          bool &Result) const {
340   const coff_section *sec = toSec(Sec);
341   Result = sec->Characteristics & COFF::IMAGE_SCN_CNT_CODE;
342   return object_error::success;
343 }
344 
345 error_code COFFObjectFile::isSectionData(DataRefImpl Sec,
346                                          bool &Result) const {
347   const coff_section *sec = toSec(Sec);
348   Result = sec->Characteristics & COFF::IMAGE_SCN_CNT_INITIALIZED_DATA;
349   return object_error::success;
350 }
351 
352 error_code COFFObjectFile::isSectionBSS(DataRefImpl Sec,
353                                         bool &Result) const {
354   const coff_section *sec = toSec(Sec);
355   Result = sec->Characteristics & COFF::IMAGE_SCN_CNT_UNINITIALIZED_DATA;
356   return object_error::success;
357 }
358 
359 error_code COFFObjectFile::sectionContainsSymbol(DataRefImpl Sec,
360                                                  DataRefImpl Symb,
361                                                  bool &Result) const {
362   // FIXME: Unimplemented.
363   Result = false;
364   return object_error::success;
365 }
366 
367 COFFObjectFile::COFFObjectFile(MemoryBuffer *Object, error_code &ec)
368   : ObjectFile(Binary::isCOFF, Object, ec) {
369   // Check that we at least have enough room for a header.
370   if (!checkSize(Data, ec, sizeof(coff_file_header))) return;
371 
372   // The actual starting location of the COFF header in the file. This can be
373   // non-zero in PE/COFF files.
374   uint64_t HeaderStart = 0;
375 
376   // Check if this is a PE/COFF file.
377   if (base()[0] == 0x4d && base()[1] == 0x5a) {
378     // PE/COFF, seek through MS-DOS compatibility stub and 4-byte
379     // PE signature to find 'normal' COFF header.
380     if (!checkSize(Data, ec, 0x3c + 8)) return;
381     HeaderStart += *reinterpret_cast<const ulittle32_t *>(base() + 0x3c);
382     // Check the PE header. ("PE\0\0")
383     if (std::memcmp(base() + HeaderStart, "PE\0\0", 4) != 0) {
384       ec = object_error::parse_failed;
385       return;
386     }
387     HeaderStart += 4; // Skip the PE Header.
388   }
389 
390   Header = reinterpret_cast<const coff_file_header *>(base() + HeaderStart);
391   if (!checkAddr(Data, ec, uintptr_t(Header), sizeof(coff_file_header)))
392     return;
393 
394   SectionTable =
395     reinterpret_cast<const coff_section *>( base()
396                                           + HeaderStart
397                                           + sizeof(coff_file_header)
398                                           + Header->SizeOfOptionalHeader);
399   if (!checkAddr(Data, ec, uintptr_t(SectionTable),
400                  Header->NumberOfSections * sizeof(coff_section)))
401     return;
402 
403   SymbolTable =
404     reinterpret_cast<const coff_symbol *>(base()
405                                           + Header->PointerToSymbolTable);
406   if (!checkAddr(Data, ec, uintptr_t(SymbolTable),
407                  Header->NumberOfSymbols * sizeof(coff_symbol)))
408     return;
409 
410   // Find string table.
411   StringTable = reinterpret_cast<const char *>(base())
412                 + Header->PointerToSymbolTable
413                 + Header->NumberOfSymbols * sizeof(coff_symbol);
414   if (!checkAddr(Data, ec, uintptr_t(StringTable), sizeof(ulittle32_t)))
415     return;
416 
417   StringTableSize = *reinterpret_cast<const ulittle32_t *>(StringTable);
418   if (!checkAddr(Data, ec, uintptr_t(StringTable), StringTableSize))
419     return;
420   // Check that the string table is null terminated if has any in it.
421   if (StringTableSize < 4
422       || (StringTableSize > 4 && StringTable[StringTableSize - 1] != 0)) {
423     ec = object_error::parse_failed;
424     return;
425   }
426 
427   ec = object_error::success;
428 }
429 
430 ObjectFile::symbol_iterator COFFObjectFile::begin_symbols() const {
431   DataRefImpl ret;
432   std::memset(&ret, 0, sizeof(DataRefImpl));
433   ret.p = reinterpret_cast<intptr_t>(SymbolTable);
434   return symbol_iterator(SymbolRef(ret, this));
435 }
436 
437 ObjectFile::symbol_iterator COFFObjectFile::end_symbols() const {
438   // The symbol table ends where the string table begins.
439   DataRefImpl ret;
440   std::memset(&ret, 0, sizeof(DataRefImpl));
441   ret.p = reinterpret_cast<intptr_t>(StringTable);
442   return symbol_iterator(SymbolRef(ret, this));
443 }
444 
445 ObjectFile::section_iterator COFFObjectFile::begin_sections() const {
446   DataRefImpl ret;
447   std::memset(&ret, 0, sizeof(DataRefImpl));
448   ret.p = reinterpret_cast<intptr_t>(SectionTable);
449   return section_iterator(SectionRef(ret, this));
450 }
451 
452 ObjectFile::section_iterator COFFObjectFile::end_sections() const {
453   DataRefImpl ret;
454   std::memset(&ret, 0, sizeof(DataRefImpl));
455   ret.p = reinterpret_cast<intptr_t>(SectionTable + Header->NumberOfSections);
456   return section_iterator(SectionRef(ret, this));
457 }
458 
459 uint8_t COFFObjectFile::getBytesInAddress() const {
460   return getArch() == Triple::x86_64 ? 8 : 4;
461 }
462 
463 StringRef COFFObjectFile::getFileFormatName() const {
464   switch(Header->Machine) {
465   case COFF::IMAGE_FILE_MACHINE_I386:
466     return "COFF-i386";
467   case COFF::IMAGE_FILE_MACHINE_AMD64:
468     return "COFF-x86-64";
469   default:
470     return "COFF-<unknown arch>";
471   }
472 }
473 
474 unsigned COFFObjectFile::getArch() const {
475   switch(Header->Machine) {
476   case COFF::IMAGE_FILE_MACHINE_I386:
477     return Triple::x86;
478   case COFF::IMAGE_FILE_MACHINE_AMD64:
479     return Triple::x86_64;
480   default:
481     return Triple::UnknownArch;
482   }
483 }
484 
485 error_code COFFObjectFile::getSection(int32_t index,
486                                       const coff_section *&Result) const {
487   // Check for special index values.
488   if (index == COFF::IMAGE_SYM_UNDEFINED ||
489       index == COFF::IMAGE_SYM_ABSOLUTE ||
490       index == COFF::IMAGE_SYM_DEBUG)
491     Result = NULL;
492   else if (index > 0 && index <= Header->NumberOfSections)
493     // We already verified the section table data, so no need to check again.
494     Result = SectionTable + (index - 1);
495   else
496     return object_error::parse_failed;
497   return object_error::success;
498 }
499 
500 error_code COFFObjectFile::getString(uint32_t offset,
501                                      StringRef &Result) const {
502   if (StringTableSize <= 4)
503     // Tried to get a string from an empty string table.
504     return object_error::parse_failed;
505   if (offset >= StringTableSize)
506     return object_error::unexpected_eof;
507   Result = StringRef(StringTable + offset);
508   return object_error::success;
509 }
510 
511 const coff_relocation *COFFObjectFile::toRel(DataRefImpl Rel) const {
512   assert(Rel.d.b < Header->NumberOfSections && "Section index out of range!");
513   const coff_section *Sect = NULL;
514   getSection(Rel.d.b, Sect);
515   assert(Rel.d.a < Sect->NumberOfRelocations && "Relocation index out of range!");
516   return
517     reinterpret_cast<const coff_relocation*>(base() +
518                                              Sect->PointerToRelocations) +
519                                              Rel.d.a;
520 }
521 error_code COFFObjectFile::getRelocationNext(DataRefImpl Rel,
522                                              RelocationRef &Res) const {
523   const coff_section *Sect = NULL;
524   if (error_code ec = getSection(Rel.d.b, Sect))
525     return ec;
526   if (++Rel.d.a >= Sect->NumberOfRelocations) {
527     Rel.d.a = 0;
528     while (++Rel.d.b < Header->NumberOfSections) {
529       const coff_section *Sect = NULL;
530       getSection(Rel.d.b, Sect);
531       if (Sect->NumberOfRelocations > 0)
532         break;
533     }
534   }
535   Res = RelocationRef(Rel, this);
536   return object_error::success;
537 }
538 error_code COFFObjectFile::getRelocationAddress(DataRefImpl Rel,
539                                                 uint64_t &Res) const {
540   const coff_section *Sect = NULL;
541   if (error_code ec = getSection(Rel.d.b, Sect))
542     return ec;
543   const coff_relocation* R = toRel(Rel);
544   Res = reinterpret_cast<uintptr_t>(base() +
545                                     Sect->PointerToRawData +
546                                     R->VirtualAddress);
547   return object_error::success;
548 }
549 error_code COFFObjectFile::getRelocationSymbol(DataRefImpl Rel,
550                                                SymbolRef &Res) const {
551   const coff_relocation* R = toRel(Rel);
552   DataRefImpl Symb;
553   Symb.p = reinterpret_cast<uintptr_t>(SymbolTable + R->SymbolTableIndex);
554   Res = SymbolRef(Symb, this);
555   return object_error::success;
556 }
557 error_code COFFObjectFile::getRelocationType(DataRefImpl Rel,
558                                              uint32_t &Res) const {
559   const coff_relocation* R = toRel(Rel);
560   Res = R->Type;
561   return object_error::success;
562 }
563 error_code COFFObjectFile::getRelocationAdditionalInfo(DataRefImpl Rel,
564                                                        int64_t &Res) const {
565   Res = 0;
566   return object_error::success;
567 }
568 ObjectFile::relocation_iterator COFFObjectFile::begin_relocations() const {
569   DataRefImpl ret;
570   ret.d.a = 0;
571   ret.d.b = 1;
572   return relocation_iterator(RelocationRef(ret, this));
573 }
574 ObjectFile::relocation_iterator COFFObjectFile::end_relocations() const {
575   DataRefImpl ret;
576   ret.d.a = 0;
577   ret.d.b = Header->NumberOfSections;
578   return relocation_iterator(RelocationRef(ret, this));
579 }
580 
581 
582 namespace llvm {
583 
584   ObjectFile *ObjectFile::createCOFFObjectFile(MemoryBuffer *Object) {
585     error_code ec;
586     return new COFFObjectFile(Object, ec);
587   }
588 
589 } // end namespace llvm
590