courgette/disassembler_win32_x86.cc

   1 // Copyright (c) 2011 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "courgette/disassembler_win32_x86.h"
   6
   7 #include <algorithm>
   8 #include <string>
   9 #include <vector>
  10
  11 #include "base/basictypes.h"
  12 #include "base/logging.h"
  13
  14 #include "courgette/assembly_program.h"
  15 #include "courgette/courgette.h"
  16 #include "courgette/encoded_program.h"
  17 #include "courgette/rel32_finder_win32_x86.h"
  18
  19 namespace courgette {
  20
  21 DisassemblerWin32X86::DisassemblerWin32X86(const void* start, size_t length)
  22   : Disassembler(start, length),
  23     incomplete_disassembly_(false),
  24     is_PE32_plus_(false),
  25     optional_header_(NULL),
  26     size_of_optional_header_(0),
  27     offset_of_data_directories_(0),
  28     machine_type_(0),
  29     number_of_sections_(0),
  30     sections_(NULL),
  31     has_text_section_(false),
  32     size_of_code_(0),
  33     size_of_initialized_data_(0),
  34     size_of_uninitialized_data_(0),
  35     base_of_code_(0),
  36     base_of_data_(0),
  37     image_base_(0),
  38     size_of_image_(0),
  39     number_of_data_directories_(0) {
  40 }
  41
  42 // ParseHeader attempts to match up the buffer with the Windows data
  43 // structures that exist within a Windows 'Portable Executable' format file.
  44 // Returns 'true' if the buffer matches, and 'false' if the data looks
  45 // suspicious.  Rather than try to 'map' the buffer to the numerous windows
  46 // structures, we extract the information we need into the courgette::PEInfo
  47 // structure.
  48 //
  49 bool DisassemblerWin32X86::ParseHeader() {
  50   if (length() < kOffsetOfFileAddressOfNewExeHeader + 4 /*size*/)
  51     return Bad("Too small");
  52
  53   // Have 'MZ' magic for a DOS header?
  54   if (start()[0] != 'M' || start()[1] != 'Z')
  55     return Bad("Not MZ");
  56
  57   // offset from DOS header to PE header is stored in DOS header.
  58   uint32 offset = ReadU32(start(),
  59                           kOffsetOfFileAddressOfNewExeHeader);
  60
  61   if (offset >= length())
  62     return Bad("Bad offset to PE header");
  63
  64   const uint8* const pe_header = OffsetToPointer(offset);
  65   const size_t kMinPEHeaderSize = 4 /*signature*/ + kSizeOfCoffHeader;
  66   if (pe_header <= start() ||
  67       pe_header >= end() - kMinPEHeaderSize)
  68     return Bad("Bad offset to PE header");
  69
  70   if (offset % 8 != 0)
  71     return Bad("Misaligned PE header");
  72
  73   // The 'PE' header is an IMAGE_NT_HEADERS structure as defined in WINNT.H.
  74   // See http://msdn.microsoft.com/en-us/library/ms680336(VS.85).aspx
  75   //
  76   // The first field of the IMAGE_NT_HEADERS is the signature.
  77   if (!(pe_header[0] == 'P' &&
  78         pe_header[1] == 'E' &&
  79         pe_header[2] == 0 &&
  80         pe_header[3] == 0))
  81     return Bad("no PE signature");
  82
  83   // The second field of the IMAGE_NT_HEADERS is the COFF header.
  84   // The COFF header is also called an IMAGE_FILE_HEADER
  85   //   http://msdn.microsoft.com/en-us/library/ms680313(VS.85).aspx
  86   const uint8* const coff_header = pe_header + 4;
  87   machine_type_       = ReadU16(coff_header, 0);
  88   number_of_sections_ = ReadU16(coff_header, 2);
  89   size_of_optional_header_ = ReadU16(coff_header, 16);
  90
  91   // The rest of the IMAGE_NT_HEADERS is the IMAGE_OPTIONAL_HEADER(32|64)
  92   const uint8* const optional_header = coff_header + kSizeOfCoffHeader;
  93   optional_header_ = optional_header;
  94
  95   if (optional_header + size_of_optional_header_ >= end())
  96     return Bad("optional header past end of file");
  97
  98   // Check we can read the magic.
  99   if (size_of_optional_header_ < 2)
 100     return Bad("optional header no magic");
 101
 102   uint16 magic = ReadU16(optional_header, 0);
 103
 104   if (magic == kImageNtOptionalHdr32Magic) {
 105     is_PE32_plus_ = false;
 106     offset_of_data_directories_ =
 107       kOffsetOfDataDirectoryFromImageOptionalHeader32;
 108   } else if (magic == kImageNtOptionalHdr64Magic) {
 109     is_PE32_plus_ = true;
 110     offset_of_data_directories_ =
 111       kOffsetOfDataDirectoryFromImageOptionalHeader64;
 112   } else {
 113     return Bad("unrecognized magic");
 114   }
 115
 116   // Check that we can read the rest of the the fixed fields.  Data directories
 117   // directly follow the fixed fields of the IMAGE_OPTIONAL_HEADER.
 118   if (size_of_optional_header_ < offset_of_data_directories_)
 119     return Bad("optional header too short");
 120
 121   // The optional header is either an IMAGE_OPTIONAL_HEADER32 or
 122   // IMAGE_OPTIONAL_HEADER64
 123   // http://msdn.microsoft.com/en-us/library/ms680339(VS.85).aspx
 124   //
 125   // Copy the fields we care about.
 126   size_of_code_               = ReadU32(optional_header, 4);
 127   size_of_initialized_data_   = ReadU32(optional_header, 8);
 128   size_of_uninitialized_data_ = ReadU32(optional_header, 12);
 129   base_of_code_               = ReadU32(optional_header, 20);
 130   if (is_PE32_plus_) {
 131     base_of_data_ = 0;
 132     image_base_  = ReadU64(optional_header, 24);
 133   } else {
 134     base_of_data_ = ReadU32(optional_header, 24);
 135     image_base_   = ReadU32(optional_header, 28);
 136   }
 137   size_of_image_ = ReadU32(optional_header, 56);
 138   number_of_data_directories_ =
 139     ReadU32(optional_header, (is_PE32_plus_ ? 108 : 92));
 140
 141   if (size_of_code_ >= length() ||
 142       size_of_initialized_data_ >= length() ||
 143       size_of_code_ + size_of_initialized_data_ >= length()) {
 144     // This validation fires on some perfectly fine executables.
 145     //  return Bad("code or initialized data too big");
 146   }
 147
 148   // TODO(sra): we can probably get rid of most of the data directories.
 149   bool b = true;
 150   // 'b &= ...' could be short circuit 'b = b && ...' but it is not necessary
 151   // for correctness and it compiles smaller this way.
 152   b &= ReadDataDirectory(0, &export_table_);
 153   b &= ReadDataDirectory(1, &import_table_);
 154   b &= ReadDataDirectory(2, &resource_table_);
 155   b &= ReadDataDirectory(3, &exception_table_);
 156   b &= ReadDataDirectory(5, &base_relocation_table_);
 157   b &= ReadDataDirectory(11, &bound_import_table_);
 158   b &= ReadDataDirectory(12, &import_address_table_);
 159   b &= ReadDataDirectory(13, &delay_import_descriptor_);
 160   b &= ReadDataDirectory(14, &clr_runtime_header_);
 161   if (!b) {
 162     return Bad("malformed data directory");
 163   }
 164
 165   // Sections follow the optional header.
 166   sections_ =
 167       reinterpret_cast<const Section*>(optional_header +
 168                                        size_of_optional_header_);
 169   size_t detected_length = 0;
 170
 171   for (int i = 0;  i < number_of_sections_;  ++i) {
 172     const Section* section = &sections_[i];
 173
 174     // TODO(sra): consider using the 'characteristics' field of the section
 175     // header to see if the section contains instructions.
 176     if (memcmp(section->name, ".text", 6) == 0)
 177       has_text_section_ = true;
 178
 179     uint32 section_end =
 180         section->file_offset_of_raw_data + section->size_of_raw_data;
 181     if (section_end > detected_length)
 182       detected_length = section_end;
 183   }
 184
 185   // Pretend our in-memory copy is only as long as our detected length.
 186   ReduceLength(detected_length);
 187
 188   if (!is_32bit()) {
 189     return Bad("64 bit executables are not supported by this disassembler");
 190   }
 191
 192   if (!has_text_section()) {
 193     return Bad("Resource-only executables are not yet supported");
 194   }
 195
 196   return Good();
 197 }
 198
 199 bool DisassemblerWin32X86::Disassemble(AssemblyProgram* target) {
 200   if (!ok())
 201     return false;
 202
 203   target->set_image_base(image_base());
 204
 205   if (!ParseAbs32Relocs())
 206     return false;
 207
 208   ParseRel32RelocsFromSections();
 209
 210   if (!ParseFile(target))
 211     return false;
 212
 213   target->DefaultAssignIndexes();
 214
 215   return true;
 216 }
 217
 218 ////////////////////////////////////////////////////////////////////////////////
 219
 220 bool DisassemblerWin32X86::ParseRelocs(std::vector<RVA> *relocs) {
 221   relocs->clear();
 222
 223   size_t relocs_size = base_relocation_table_.size_;
 224   if (relocs_size == 0)
 225     return true;
 226
 227   // The format of the base relocation table is a sequence of variable sized
 228   // IMAGE_BASE_RELOCATION blocks.  Search for
 229   //   "The format of the base relocation data is somewhat quirky"
 230   // at http://msdn.microsoft.com/en-us/library/ms809762.aspx
 231
 232   const uint8* relocs_start = RVAToPointer(base_relocation_table_.address_);
 233   const uint8* relocs_end = relocs_start + relocs_size;
 234
 235   // Make sure entire base relocation table is within the buffer.
 236   if (relocs_start < start() ||
 237       relocs_start >= end() ||
 238       relocs_end <= start() ||
 239       relocs_end > end()) {
 240     return Bad(".relocs outside image");
 241   }
 242
 243   const uint8* block = relocs_start;
 244
 245   // Walk the variable sized blocks.
 246   while (block + 8 < relocs_end) {
 247     RVA page_rva = ReadU32(block, 0);
 248     uint32 size = ReadU32(block, 4);
 249     if (size < 8 ||        // Size includes header ...
 250         size % 4  !=  0)   // ... and is word aligned.
 251       return Bad("unreasonable relocs block");
 252
 253     const uint8* end_entries = block + size;
 254
 255     if (end_entries <= block ||
 256         end_entries <= start() ||
 257         end_entries > end())
 258       return Bad(".relocs block outside image");
 259
 260     // Walk through the two-byte entries.
 261     for (const uint8* p = block + 8;  p < end_entries;  p += 2) {
 262       uint16 entry = ReadU16(p, 0);
 263       int type = entry >> 12;
 264       int offset = entry & 0xFFF;
 265
 266       RVA rva = page_rva + offset;
 267       // Skip the relocs that live outside of the image. It might be the case
 268       // if a reloc is relative to a register, e.g.:
 269       //     mov    ecx,dword ptr [eax+044D5888h]
 270       uint32 target_address = Read32LittleEndian(RVAToPointer(rva));
 271       if (target_address < image_base_ ||
 272           target_address > (image_base_ + size_of_image_)) {
 273         continue;
 274       }
 275       if (type == 3) {         // IMAGE_REL_BASED_HIGHLOW
 276         relocs->push_back(rva);
 277       } else if (type == 0) {  // IMAGE_REL_BASED_ABSOLUTE
 278         // Ignore, used as padding.
 279       } else {
 280         // Does not occur in Windows x86 executables.
 281         return Bad("unknown type of reloc");
 282       }
 283     }
 284
 285     block += size;
 286   }
 287
 288   std::sort(relocs->begin(), relocs->end());
 289
 290   return true;
 291 }
 292
 293 const Section* DisassemblerWin32X86::RVAToSection(RVA rva) const {
 294   for (int i = 0; i < number_of_sections_; i++) {
 295     const Section* section = &sections_[i];
 296     uint32 offset = rva - section->virtual_address;
 297     if (offset < section->virtual_size) {
 298       return section;
 299     }
 300   }
 301   return NULL;
 302 }
 303
 304 int DisassemblerWin32X86::RVAToFileOffset(RVA rva) const {
 305   const Section* section = RVAToSection(rva);
 306   if (section) {
 307     uint32 offset = rva - section->virtual_address;
 308     if (offset < section->size_of_raw_data) {
 309       return section->file_offset_of_raw_data + offset;
 310     } else {
 311       return kNoOffset;  // In section but not in file (e.g. uninit data).
 312     }
 313   }
 314
 315   // Small RVA values point into the file header in the loaded image.
 316   // RVA 0 is the module load address which Windows uses as the module handle.
 317   // RVA 2 sometimes occurs, I'm not sure what it is, but it would map into the
 318   // DOS header.
 319   if (rva == 0 || rva == 2)
 320     return rva;
 321
 322   NOTREACHED();
 323   return kNoOffset;
 324 }
 325
 326 const uint8* DisassemblerWin32X86::RVAToPointer(RVA rva) const {
 327   int file_offset = RVAToFileOffset(rva);
 328   if (file_offset == kNoOffset)
 329     return NULL;
 330   else
 331     return OffsetToPointer(file_offset);
 332 }
 333
 334 std::string DisassemblerWin32X86::SectionName(const Section* section) {
 335   if (section == NULL)
 336     return "<none>";
 337   char name[9];
 338   memcpy(name, section->name, 8);
 339   name[8] = '\0';  // Ensure termination.
 340   return name;
 341 }
 342
 343 CheckBool DisassemblerWin32X86::ParseFile(AssemblyProgram* program) {
 344   // Walk all the bytes in the file, whether or not in a section.
 345   uint32 file_offset = 0;
 346   while (file_offset < length()) {
 347     const Section* section = FindNextSection(file_offset);
 348     if (section == NULL) {
 349       // No more sections.  There should not be extra stuff following last
 350       // section.
 351       //   ParseNonSectionFileRegion(file_offset, pe_info().length(), program);
 352       break;
 353     }
 354     if (file_offset < section->file_offset_of_raw_data) {
 355       uint32 section_start_offset = section->file_offset_of_raw_data;
 356       if(!ParseNonSectionFileRegion(file_offset, section_start_offset,
 357                                     program))
 358         return false;
 359
 360       file_offset = section_start_offset;
 361     }
 362     uint32 end = file_offset + section->size_of_raw_data;
 363     if (!ParseFileRegion(section, file_offset, end, program))
 364       return false;
 365     file_offset = end;
 366   }
 367
 368 #if COURGETTE_HISTOGRAM_TARGETS
 369   HistogramTargets("abs32 relocs", abs32_target_rvas_);
 370   HistogramTargets("rel32 relocs", rel32_target_rvas_);
 371 #endif
 372
 373   return true;
 374 }
 375
 376 bool DisassemblerWin32X86::ParseAbs32Relocs() {
 377   abs32_locations_.clear();
 378   if (!ParseRelocs(&abs32_locations_))
 379     return false;
 380
 381 #if COURGETTE_HISTOGRAM_TARGETS
 382   for (size_t i = 0;  i < abs32_locations_.size(); ++i) {
 383     RVA rva = abs32_locations_[i];
 384     // The 4 bytes at the relocation are a reference to some address.
 385     uint32 target_address = Read32LittleEndian(RVAToPointer(rva));
 386     ++abs32_target_rvas_[target_address - image_base()];
 387   }
 388 #endif
 389   return true;
 390 }
 391
 392 void DisassemblerWin32X86::ParseRel32RelocsFromSections() {
 393   uint32 file_offset = 0;
 394   while (file_offset < length()) {
 395     const Section* section = FindNextSection(file_offset);
 396     if (section == NULL)
 397       break;
 398     if (file_offset < section->file_offset_of_raw_data)
 399       file_offset = section->file_offset_of_raw_data;
 400     ParseRel32RelocsFromSection(section);
 401     file_offset += section->size_of_raw_data;
 402   }
 403   std::sort(rel32_locations_.begin(), rel32_locations_.end());
 404
 405 #if COURGETTE_HISTOGRAM_TARGETS
 406   VLOG(1) << "abs32_locations_ " << abs32_locations_.size()
 407           << "\nrel32_locations_ " << rel32_locations_.size()
 408           << "\nabs32_target_rvas_ " << abs32_target_rvas_.size()
 409           << "\nrel32_target_rvas_ " << rel32_target_rvas_.size();
 410
 411   int common = 0;
 412   std::map<RVA, int>::iterator abs32_iter = abs32_target_rvas_.begin();
 413   std::map<RVA, int>::iterator rel32_iter = rel32_target_rvas_.begin();
 414   while (abs32_iter != abs32_target_rvas_.end() &&
 415          rel32_iter != rel32_target_rvas_.end()) {
 416     if (abs32_iter->first < rel32_iter->first)
 417       ++abs32_iter;
 418     else if (rel32_iter->first < abs32_iter->first)
 419       ++rel32_iter;
 420     else {
 421       ++common;
 422       ++abs32_iter;
 423       ++rel32_iter;
 424     }
 425   }
 426   VLOG(1) << "common " << common;
 427 #endif
 428 }
 429
 430 void DisassemblerWin32X86::ParseRel32RelocsFromSection(const Section* section) {
 431   // TODO(sra): use characteristic.
 432   bool isCode = strcmp(section->name, ".text") == 0;
 433   if (!isCode)
 434     return;
 435
 436   uint32 start_file_offset = section->file_offset_of_raw_data;
 437   uint32 end_file_offset = start_file_offset + section->size_of_raw_data;
 438
 439   const uint8* start_pointer = OffsetToPointer(start_file_offset);
 440   const uint8* end_pointer = OffsetToPointer(end_file_offset);
 441
 442   RVA start_rva = FileOffsetToRVA(start_file_offset);
 443   RVA end_rva = start_rva + section->virtual_size;
 444
 445   Rel32FinderWin32X86_Basic finder(
 446       base_relocation_table().address_,
 447       base_relocation_table().address_ + base_relocation_table().size_,
 448       size_of_image_);
 449   finder.Find(start_pointer, end_pointer, start_rva, end_rva, abs32_locations_);
 450   finder.SwapRel32Locations(&rel32_locations_);
 451
 452 #if COURGETTE_HISTOGRAM_TARGETS
 453   DCHECK(rel32_target_rvas_.empty());
 454   finder.SwapRel32TargetRVAs(&rel32_target_rvas_);
 455 #endif
 456 }
 457
 458 CheckBool DisassemblerWin32X86::ParseNonSectionFileRegion(
 459     uint32 start_file_offset,
 460     uint32 end_file_offset,
 461     AssemblyProgram* program) {
 462   if (incomplete_disassembly_)
 463     return true;
 464
 465   if (end_file_offset > start_file_offset) {
 466     if (!program->EmitBytesInstruction(OffsetToPointer(start_file_offset),
 467                                        end_file_offset - start_file_offset)) {
 468       return false;
 469     }
 470   }
 471
 472   return true;
 473 }
 474
 475 CheckBool DisassemblerWin32X86::ParseFileRegion(
 476     const Section* section,
 477     uint32 start_file_offset, uint32 end_file_offset,
 478     AssemblyProgram* program) {
 479   RVA relocs_start_rva = base_relocation_table().address_;
 480
 481   const uint8* start_pointer = OffsetToPointer(start_file_offset);
 482   const uint8* end_pointer = OffsetToPointer(end_file_offset);
 483
 484   RVA start_rva = FileOffsetToRVA(start_file_offset);
 485   RVA end_rva = start_rva + section->virtual_size;
 486
 487   // Quick way to convert from Pointer to RVA within a single Section is to
 488   // subtract 'pointer_to_rva'.
 489   const uint8* const adjust_pointer_to_rva = start_pointer - start_rva;
 490
 491   std::vector<RVA>::iterator rel32_pos = rel32_locations_.begin();
 492   std::vector<RVA>::iterator abs32_pos = abs32_locations_.begin();
 493
 494   if (!program->EmitOriginInstruction(start_rva))
 495     return false;
 496
 497   const uint8* p = start_pointer;
 498
 499   while (p < end_pointer) {
 500     RVA current_rva = static_cast<RVA>(p - adjust_pointer_to_rva);
 501
 502     // The base relocation table is usually in the .relocs section, but it could
 503     // actually be anywhere.  Make sure we skip it because we will regenerate it
 504     // during assembly.
 505     if (current_rva == relocs_start_rva) {
 506       if (!program->EmitPeRelocsInstruction())
 507         return false;
 508       uint32 relocs_size = base_relocation_table().size_;
 509       if (relocs_size) {
 510         p += relocs_size;
 511         continue;
 512       }
 513     }
 514
 515     while (abs32_pos != abs32_locations_.end() && *abs32_pos < current_rva)
 516       ++abs32_pos;
 517
 518     if (abs32_pos != abs32_locations_.end() && *abs32_pos == current_rva) {
 519       uint32 target_address = Read32LittleEndian(p);
 520       RVA target_rva = target_address - image_base();
 521       // TODO(sra): target could be Label+offset.  It is not clear how to guess
 522       // which it might be.  We assume offset==0.
 523       if (!program->EmitAbs32(program->FindOrMakeAbs32Label(target_rva)))
 524         return false;
 525       p += 4;
 526       continue;
 527     }
 528
 529     while (rel32_pos != rel32_locations_.end() && *rel32_pos < current_rva)
 530       ++rel32_pos;
 531
 532     if (rel32_pos != rel32_locations_.end() && *rel32_pos == current_rva) {
 533       RVA target_rva = current_rva + 4 + Read32LittleEndian(p);
 534       if (!program->EmitRel32(program->FindOrMakeRel32Label(target_rva)))
 535         return false;
 536       p += 4;
 537       continue;
 538     }
 539
 540     if (incomplete_disassembly_) {
 541       if ((abs32_pos == abs32_locations_.end() || end_rva <= *abs32_pos) &&
 542           (rel32_pos == rel32_locations_.end() || end_rva <= *rel32_pos) &&
 543           (end_rva <= relocs_start_rva || current_rva >= relocs_start_rva)) {
 544         // No more relocs in this section, don't bother encoding bytes.
 545         break;
 546       }
 547     }
 548
 549     if (!program->EmitByteInstruction(*p))
 550       return false;
 551     p += 1;
 552   }
 553
 554   return true;
 555 }
 556
 557 #if COURGETTE_HISTOGRAM_TARGETS
 558 // Histogram is printed to std::cout.  It is purely for debugging the algorithm
 559 // and is only enabled manually in 'exploration' builds.  I don't want to add
 560 // command-line configuration for this feature because this code has to be
 561 // small, which means compiled-out.
 562 void DisassemblerWin32X86::HistogramTargets(const char* kind,
 563                                             const std::map<RVA, int>& map) {
 564   int total = 0;
 565   std::map<int, std::vector<RVA> > h;
 566   for (std::map<RVA, int>::const_iterator p = map.begin();
 567        p != map.end();
 568        ++p) {
 569     h[p->second].push_back(p->first);
 570     total += p->second;
 571   }
 572
 573   std::cout << total << " " << kind << " to "
 574             << map.size() << " unique targets" << std::endl;
 575
 576   std::cout << "indegree: #targets-with-indegree (example)" << std::endl;
 577   const int kFirstN = 15;
 578   bool someSkipped = false;
 579   int index = 0;
 580   for (std::map<int, std::vector<RVA> >::reverse_iterator p = h.rbegin();
 581        p != h.rend();
 582        ++p) {
 583     ++index;
 584     if (index <= kFirstN || p->first <= 3) {
 585       if (someSkipped) {
 586         std::cout << "..." << std::endl;
 587       }
 588       size_t count = p->second.size();
 589       std::cout << std::dec << p->first << ": " << count;
 590       if (count <= 2) {
 591         for (size_t i = 0;  i < count;  ++i)
 592           std::cout << "  " << DescribeRVA(p->second[i]);
 593       }
 594       std::cout << std::endl;
 595       someSkipped = false;
 596     } else {
 597       someSkipped = true;
 598     }
 599   }
 600 }
 601 #endif  // COURGETTE_HISTOGRAM_TARGETS
 602
 603
 604 // DescribeRVA is for debugging only.  I would put it under #ifdef DEBUG except
 605 // that during development I'm finding I need to call it when compiled in
 606 // Release mode.  Hence:
 607 // TODO(sra): make this compile only for debug mode.
 608 std::string DisassemblerWin32X86::DescribeRVA(RVA rva) const {
 609   const Section* section = RVAToSection(rva);
 610   std::ostringstream s;
 611   s << std::hex << rva;
 612   if (section) {
 613     s << " (";
 614     s << SectionName(section) << "+"
 615       << std::hex << (rva - section->virtual_address)
 616       << ")";
 617   }
 618   return s.str();
 619 }
 620
 621 const Section* DisassemblerWin32X86::FindNextSection(uint32 fileOffset) const {
 622   const Section* best = 0;
 623   for (int i = 0; i < number_of_sections_; i++) {
 624     const Section* section = &sections_[i];
 625     if (section->size_of_raw_data > 0) {  // i.e. has data in file.
 626       if (fileOffset <= section->file_offset_of_raw_data) {
 627         if (best == 0 ||
 628             section->file_offset_of_raw_data < best->file_offset_of_raw_data) {
 629           best = section;
 630         }
 631       }
 632     }
 633   }
 634   return best;
 635 }
 636
 637 RVA DisassemblerWin32X86::FileOffsetToRVA(uint32 file_offset) const {
 638   for (int i = 0; i < number_of_sections_; i++) {
 639     const Section* section = &sections_[i];
 640     uint32 offset = file_offset - section->file_offset_of_raw_data;
 641     if (offset < section->size_of_raw_data) {
 642       return section->virtual_address + offset;
 643     }
 644   }
 645   return 0;
 646 }
 647
 648 bool DisassemblerWin32X86::ReadDataDirectory(
 649     int index,
 650     ImageDataDirectory* directory) {
 651
 652   if (index < number_of_data_directories_) {
 653     size_t offset = index * 8 + offset_of_data_directories_;
 654     if (offset >= size_of_optional_header_)
 655       return Bad("number of data directories inconsistent");
 656     const uint8* data_directory = optional_header_ + offset;
 657     if (data_directory < start() ||
 658         data_directory + 8 >= end())
 659       return Bad("data directory outside image");
 660     RVA rva = ReadU32(data_directory, 0);
 661     size_t size  = ReadU32(data_directory, 4);
 662     if (size > size_of_image_)
 663       return Bad("data directory size too big");
 664
 665     // TODO(sra): validate RVA.
 666     directory->address_ = rva;
 667     directory->size_ = static_cast<uint32>(size);
 668     return true;
 669   } else {
 670     directory->address_ = 0;
 671     directory->size_ = 0;
 672     return true;
 673   }
 674 }
 675
 676 }  // namespace courgette