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
  18 // COURGETTE_HISTOGRAM_TARGETS prints out a histogram of how frequently
  19 // different target addresses are referenced.  Purely for debugging.
  20 #define COURGETTE_HISTOGRAM_TARGETS 0
  21
  22 namespace courgette {
  23
  24 DisassemblerWin32X86::DisassemblerWin32X86(const void* start, size_t length)
  25   : Disassembler(start, length),
  26     incomplete_disassembly_(false),
  27     is_PE32_plus_(false),
  28     optional_header_(NULL),
  29     size_of_optional_header_(0),
  30     offset_of_data_directories_(0),
  31     machine_type_(0),
  32     number_of_sections_(0),
  33     sections_(NULL),
  34     has_text_section_(false),
  35     size_of_code_(0),
  36     size_of_initialized_data_(0),
  37     size_of_uninitialized_data_(0),
  38     base_of_code_(0),
  39     base_of_data_(0),
  40     image_base_(0),
  41     size_of_image_(0),
  42     number_of_data_directories_(0) {
  43 }
  44
  45 // ParseHeader attempts to match up the buffer with the Windows data
  46 // structures that exist within a Windows 'Portable Executable' format file.
  47 // Returns 'true' if the buffer matches, and 'false' if the data looks
  48 // suspicious.  Rather than try to 'map' the buffer to the numerous windows
  49 // structures, we extract the information we need into the courgette::PEInfo
  50 // structure.
  51 //
  52 bool DisassemblerWin32X86::ParseHeader() {
  53   if (length() < kOffsetOfFileAddressOfNewExeHeader + 4 /*size*/)
  54     return Bad("Too small");
  55
  56   // Have 'MZ' magic for a DOS header?
  57   if (start()[0] != 'M' || start()[1] != 'Z')
  58     return Bad("Not MZ");
  59
  60   // offset from DOS header to PE header is stored in DOS header.
  61   uint32 offset = ReadU32(start(),
  62                           kOffsetOfFileAddressOfNewExeHeader);
  63
  64   if (offset >= length())
  65     return Bad("Bad offset to PE header");
  66
  67   const uint8* const pe_header = OffsetToPointer(offset);
  68   const size_t kMinPEHeaderSize = 4 /*signature*/ + kSizeOfCoffHeader;
  69   if (pe_header <= start() ||
  70       pe_header >= end() - kMinPEHeaderSize)
  71     return Bad("Bad offset to PE header");
  72
  73   if (offset % 8 != 0)
  74     return Bad("Misaligned PE header");
  75
  76   // The 'PE' header is an IMAGE_NT_HEADERS structure as defined in WINNT.H.
  77   // See http://msdn.microsoft.com/en-us/library/ms680336(VS.85).aspx
  78   //
  79   // The first field of the IMAGE_NT_HEADERS is the signature.
  80   if (!(pe_header[0] == 'P' &&
  81         pe_header[1] == 'E' &&
  82         pe_header[2] == 0 &&
  83         pe_header[3] == 0))
  84     return Bad("no PE signature");
  85
  86   // The second field of the IMAGE_NT_HEADERS is the COFF header.
  87   // The COFF header is also called an IMAGE_FILE_HEADER
  88   //   http://msdn.microsoft.com/en-us/library/ms680313(VS.85).aspx
  89   const uint8* const coff_header = pe_header + 4;
  90   machine_type_       = ReadU16(coff_header, 0);
  91   number_of_sections_ = ReadU16(coff_header, 2);
  92   size_of_optional_header_ = ReadU16(coff_header, 16);
  93
  94   // The rest of the IMAGE_NT_HEADERS is the IMAGE_OPTIONAL_HEADER(32|64)
  95   const uint8* const optional_header = coff_header + kSizeOfCoffHeader;
  96   optional_header_ = optional_header;
  97
  98   if (optional_header + size_of_optional_header_ >= end())
  99     return Bad("optional header past end of file");
 100
 101   // Check we can read the magic.
 102   if (size_of_optional_header_ < 2)
 103     return Bad("optional header no magic");
 104
 105   uint16 magic = ReadU16(optional_header, 0);
 106
 107   if (magic == kImageNtOptionalHdr32Magic) {
 108     is_PE32_plus_ = false;
 109     offset_of_data_directories_ =
 110       kOffsetOfDataDirectoryFromImageOptionalHeader32;
 111   } else if (magic == kImageNtOptionalHdr64Magic) {
 112     is_PE32_plus_ = true;
 113     offset_of_data_directories_ =
 114       kOffsetOfDataDirectoryFromImageOptionalHeader64;
 115   } else {
 116     return Bad("unrecognized magic");
 117   }
 118
 119   // Check that we can read the rest of the the fixed fields.  Data directories
 120   // directly follow the fixed fields of the IMAGE_OPTIONAL_HEADER.
 121   if (size_of_optional_header_ < offset_of_data_directories_)
 122     return Bad("optional header too short");
 123
 124   // The optional header is either an IMAGE_OPTIONAL_HEADER32 or
 125   // IMAGE_OPTIONAL_HEADER64
 126   // http://msdn.microsoft.com/en-us/library/ms680339(VS.85).aspx
 127   //
 128   // Copy the fields we care about.
 129   size_of_code_               = ReadU32(optional_header, 4);
 130   size_of_initialized_data_   = ReadU32(optional_header, 8);
 131   size_of_uninitialized_data_ = ReadU32(optional_header, 12);
 132   base_of_code_               = ReadU32(optional_header, 20);
 133   if (is_PE32_plus_) {
 134     base_of_data_ = 0;
 135     image_base_  = ReadU64(optional_header, 24);
 136   } else {
 137     base_of_data_ = ReadU32(optional_header, 24);
 138     image_base_   = ReadU32(optional_header, 28);
 139   }
 140   size_of_image_ = ReadU32(optional_header, 56);
 141   number_of_data_directories_ =
 142     ReadU32(optional_header, (is_PE32_plus_ ? 108 : 92));
 143
 144   if (size_of_code_ >= length() ||
 145       size_of_initialized_data_ >= length() ||
 146       size_of_code_ + size_of_initialized_data_ >= length()) {
 147     // This validation fires on some perfectly fine executables.
 148     //  return Bad("code or initialized data too big");
 149   }
 150
 151   // TODO(sra): we can probably get rid of most of the data directories.
 152   bool b = true;
 153   // 'b &= ...' could be short circuit 'b = b && ...' but it is not necessary
 154   // for correctness and it compiles smaller this way.
 155   b &= ReadDataDirectory(0, &export_table_);
 156   b &= ReadDataDirectory(1, &import_table_);
 157   b &= ReadDataDirectory(2, &resource_table_);
 158   b &= ReadDataDirectory(3, &exception_table_);
 159   b &= ReadDataDirectory(5, &base_relocation_table_);
 160   b &= ReadDataDirectory(11, &bound_import_table_);
 161   b &= ReadDataDirectory(12, &import_address_table_);
 162   b &= ReadDataDirectory(13, &delay_import_descriptor_);
 163   b &= ReadDataDirectory(14, &clr_runtime_header_);
 164   if (!b) {
 165     return Bad("malformed data directory");
 166   }
 167
 168   // Sections follow the optional header.
 169   sections_ =
 170       reinterpret_cast<const Section*>(optional_header +
 171                                        size_of_optional_header_);
 172   size_t detected_length = 0;
 173
 174   for (int i = 0;  i < number_of_sections_;  ++i) {
 175     const Section* section = &sections_[i];
 176
 177     // TODO(sra): consider using the 'characteristics' field of the section
 178     // header to see if the section contains instructions.
 179     if (memcmp(section->name, ".text", 6) == 0)
 180       has_text_section_ = true;
 181
 182     uint32 section_end =
 183         section->file_offset_of_raw_data + section->size_of_raw_data;
 184     if (section_end > detected_length)
 185       detected_length = section_end;
 186   }
 187
 188   // Pretend our in-memory copy is only as long as our detected length.
 189   ReduceLength(detected_length);
 190
 191   if (!is_32bit()) {
 192     return Bad("64 bit executables are not yet supported");
 193   }
 194
 195   if (!has_text_section()) {
 196     return Bad("Resource-only executables are not yet supported");
 197   }
 198
 199   return Good();
 200 }
 201
 202 bool DisassemblerWin32X86::Disassemble(AssemblyProgram* target) {
 203   if (!ok())
 204     return false;
 205
 206   target->set_image_base(image_base());
 207
 208   if (!ParseAbs32Relocs())
 209     return false;
 210
 211   ParseRel32RelocsFromSections();
 212
 213   if (!ParseFile(target))
 214     return false;
 215
 216   target->DefaultAssignIndexes();
 217
 218   return true;
 219 }
 220
 221 ////////////////////////////////////////////////////////////////////////////////
 222
 223 bool DisassemblerWin32X86::ParseRelocs(std::vector<RVA> *relocs) {
 224   relocs->clear();
 225
 226   size_t relocs_size = base_relocation_table_.size_;
 227   if (relocs_size == 0)
 228     return true;
 229
 230   // The format of the base relocation table is a sequence of variable sized
 231   // IMAGE_BASE_RELOCATION blocks.  Search for
 232   //   "The format of the base relocation data is somewhat quirky"
 233   // at http://msdn.microsoft.com/en-us/library/ms809762.aspx
 234
 235   const uint8* relocs_start = RVAToPointer(base_relocation_table_.address_);
 236   const uint8* relocs_end = relocs_start + relocs_size;
 237
 238   // Make sure entire base relocation table is within the buffer.
 239   if (relocs_start < start() ||
 240       relocs_start >= end() ||
 241       relocs_end <= start() ||
 242       relocs_end > end()) {
 243     return Bad(".relocs outside image");
 244   }
 245
 246   const uint8* block = relocs_start;
 247
 248   // Walk the variable sized blocks.
 249   while (block + 8 < relocs_end) {
 250     RVA page_rva = ReadU32(block, 0);
 251     uint32 size = ReadU32(block, 4);
 252     if (size < 8 ||        // Size includes header ...
 253         size % 4  !=  0)   // ... and is word aligned.
 254       return Bad("unreasonable relocs block");
 255
 256     const uint8* end_entries = block + size;
 257
 258     if (end_entries <= block ||
 259         end_entries <= start() ||
 260         end_entries > end())
 261       return Bad(".relocs block outside image");
 262
 263     // Walk through the two-byte entries.
 264     for (const uint8* p = block + 8;  p < end_entries;  p += 2) {
 265       uint16 entry = ReadU16(p, 0);
 266       int type = entry >> 12;
 267       int offset = entry & 0xFFF;
 268
 269       RVA rva = page_rva + offset;
 270       if (type == 3) {         // IMAGE_REL_BASED_HIGHLOW
 271         relocs->push_back(rva);
 272       } else if (type == 0) {  // IMAGE_REL_BASED_ABSOLUTE
 273         // Ignore, used as padding.
 274       } else {
 275         // Does not occur in Windows x86 executables.
 276         return Bad("unknown type of reloc");
 277       }
 278     }
 279
 280     block += size;
 281   }
 282
 283   std::sort(relocs->begin(), relocs->end());
 284
 285   return true;
 286 }
 287
 288 const Section* DisassemblerWin32X86::RVAToSection(RVA rva) const {
 289   for (int i = 0; i < number_of_sections_; i++) {
 290     const Section* section = &sections_[i];
 291     uint32 offset = rva - section->virtual_address;
 292     if (offset < section->virtual_size) {
 293       return section;
 294     }
 295   }
 296   return NULL;
 297 }
 298
 299 int DisassemblerWin32X86::RVAToFileOffset(RVA rva) const {
 300   const Section* section = RVAToSection(rva);
 301   if (section) {
 302     uint32 offset = rva - section->virtual_address;
 303     if (offset < section->size_of_raw_data) {
 304       return section->file_offset_of_raw_data + offset;
 305     } else {
 306       return kNoOffset;  // In section but not in file (e.g. uninit data).
 307     }
 308   }
 309
 310   // Small RVA values point into the file header in the loaded image.
 311   // RVA 0 is the module load address which Windows uses as the module handle.
 312   // RVA 2 sometimes occurs, I'm not sure what it is, but it would map into the
 313   // DOS header.
 314   if (rva == 0 || rva == 2)
 315     return rva;
 316
 317   NOTREACHED();
 318   return kNoOffset;
 319 }
 320
 321 const uint8* DisassemblerWin32X86::RVAToPointer(RVA rva) const {
 322   int file_offset = RVAToFileOffset(rva);
 323   if (file_offset == kNoOffset)
 324     return NULL;
 325   else
 326     return OffsetToPointer(file_offset);
 327 }
 328
 329 std::string DisassemblerWin32X86::SectionName(const Section* section) {
 330   if (section == NULL)
 331     return "<none>";
 332   char name[9];
 333   memcpy(name, section->name, 8);
 334   name[8] = '\0';  // Ensure termination.
 335   return name;
 336 }
 337
 338 CheckBool DisassemblerWin32X86::ParseFile(AssemblyProgram* program) {
 339   // Walk all the bytes in the file, whether or not in a section.
 340   uint32 file_offset = 0;
 341   while (file_offset < length()) {
 342     const Section* section = FindNextSection(file_offset);
 343     if (section == NULL) {
 344       // No more sections.  There should not be extra stuff following last
 345       // section.
 346       //   ParseNonSectionFileRegion(file_offset, pe_info().length(), program);
 347       break;
 348     }
 349     if (file_offset < section->file_offset_of_raw_data) {
 350       uint32 section_start_offset = section->file_offset_of_raw_data;
 351       if(!ParseNonSectionFileRegion(file_offset, section_start_offset,
 352                                     program))
 353         return false;
 354
 355       file_offset = section_start_offset;
 356     }
 357     uint32 end = file_offset + section->size_of_raw_data;
 358     if (!ParseFileRegion(section, file_offset, end, program))
 359       return false;
 360     file_offset = end;
 361   }
 362
 363 #if COURGETTE_HISTOGRAM_TARGETS
 364   HistogramTargets("abs32 relocs", abs32_target_rvas_);
 365   HistogramTargets("rel32 relocs", rel32_target_rvas_);
 366 #endif
 367
 368   return true;
 369 }
 370
 371 bool DisassemblerWin32X86::ParseAbs32Relocs() {
 372   abs32_locations_.clear();
 373   if (!ParseRelocs(&abs32_locations_))
 374     return false;
 375
 376   std::sort(abs32_locations_.begin(), abs32_locations_.end());
 377
 378 #if COURGETTE_HISTOGRAM_TARGETS
 379   for (size_t i = 0;  i < abs32_locations_.size(); ++i) {
 380     RVA rva = abs32_locations_[i];
 381     // The 4 bytes at the relocation are a reference to some address.
 382     uint32 target_address = Read32LittleEndian(RVAToPointer(rva));
 383     ++abs32_target_rvas_[target_address - image_base()];
 384   }
 385 #endif
 386   return true;
 387 }
 388
 389 void DisassemblerWin32X86::ParseRel32RelocsFromSections() {
 390   uint32 file_offset = 0;
 391   while (file_offset < length()) {
 392     const Section* section = FindNextSection(file_offset);
 393     if (section == NULL)
 394       break;
 395     if (file_offset < section->file_offset_of_raw_data)
 396       file_offset = section->file_offset_of_raw_data;
 397     ParseRel32RelocsFromSection(section);
 398     file_offset += section->size_of_raw_data;
 399   }
 400   std::sort(rel32_locations_.begin(), rel32_locations_.end());
 401
 402 #if COURGETTE_HISTOGRAM_TARGETS
 403   VLOG(1) << "abs32_locations_ " << abs32_locations_.size()
 404           << "\nrel32_locations_ " << rel32_locations_.size()
 405           << "\nabs32_target_rvas_ " << abs32_target_rvas_.size()
 406           << "\nrel32_target_rvas_ " << rel32_target_rvas_.size();
 407
 408   int common = 0;
 409   std::map<RVA, int>::iterator abs32_iter = abs32_target_rvas_.begin();
 410   std::map<RVA, int>::iterator rel32_iter = rel32_target_rvas_.begin();
 411   while (abs32_iter != abs32_target_rvas_.end() &&
 412          rel32_iter != rel32_target_rvas_.end()) {
 413     if (abs32_iter->first < rel32_iter->first)
 414       ++abs32_iter;
 415     else if (rel32_iter->first < abs32_iter->first)
 416       ++rel32_iter;
 417     else {
 418       ++common;
 419       ++abs32_iter;
 420       ++rel32_iter;
 421     }
 422   }
 423   VLOG(1) << "common " << common;
 424 #endif
 425 }
 426
 427 void DisassemblerWin32X86::ParseRel32RelocsFromSection(const Section* section) {
 428   // TODO(sra): use characteristic.
 429   bool isCode = strcmp(section->name, ".text") == 0;
 430   if (!isCode)
 431     return;
 432
 433   uint32 start_file_offset = section->file_offset_of_raw_data;
 434   uint32 end_file_offset = start_file_offset + section->size_of_raw_data;
 435   RVA relocs_start_rva = base_relocation_table().address_;
 436
 437   const uint8* start_pointer = OffsetToPointer(start_file_offset);
 438   const uint8* end_pointer = OffsetToPointer(end_file_offset);
 439
 440   RVA start_rva = FileOffsetToRVA(start_file_offset);
 441   RVA end_rva = start_rva + section->virtual_size;
 442
 443   // Quick way to convert from Pointer to RVA within a single Section is to
 444   // subtract 'pointer_to_rva'.
 445   const uint8* const adjust_pointer_to_rva = start_pointer - start_rva;
 446
 447   std::vector<RVA>::iterator abs32_pos = abs32_locations_.begin();
 448
 449   // Find the rel32 relocations.
 450   const uint8* p = start_pointer;
 451   while (p < end_pointer) {
 452     RVA current_rva = static_cast<RVA>(p - adjust_pointer_to_rva);
 453     if (current_rva == relocs_start_rva) {
 454       uint32 relocs_size = base_relocation_table().size_;
 455       if (relocs_size) {
 456         p += relocs_size;
 457         continue;
 458       }
 459     }
 460
 461     //while (abs32_pos != abs32_locations_.end() && *abs32_pos < current_rva)
 462     //  ++abs32_pos;
 463
 464     // Heuristic discovery of rel32 locations in instruction stream: are the
 465     // next few bytes the start of an instruction containing a rel32
 466     // addressing mode?
 467     const uint8* rel32 = NULL;
 468
 469     if (p + 5 <= end_pointer) {
 470       if (*p == 0xE8 || *p == 0xE9) {  // jmp rel32 and call rel32
 471         rel32 = p + 1;
 472       }
 473     }
 474     if (p + 6 <= end_pointer) {
 475       if (*p == 0x0F  &&  (*(p+1) & 0xF0) == 0x80) {  // Jcc long form
 476         if (p[1] != 0x8A && p[1] != 0x8B)  // JPE/JPO unlikely
 477           rel32 = p + 2;
 478       }
 479     }
 480     if (rel32) {
 481       RVA rel32_rva = static_cast<RVA>(rel32 - adjust_pointer_to_rva);
 482
 483       // Is there an abs32 reloc overlapping the candidate?
 484       while (abs32_pos != abs32_locations_.end() && *abs32_pos < rel32_rva - 3)
 485         ++abs32_pos;
 486       // Now: (*abs32_pos > rel32_rva - 4) i.e. the lowest addressed 4-byte
 487       // region that could overlap rel32_rva.
 488       if (abs32_pos != abs32_locations_.end()) {
 489         if (*abs32_pos < rel32_rva + 4) {
 490           // Beginning of abs32 reloc is before end of rel32 reloc so they
 491           // overlap.  Skip four bytes past the abs32 reloc.
 492           p += (*abs32_pos + 4) - current_rva;
 493           continue;
 494         }
 495       }
 496
 497       RVA target_rva = rel32_rva + 4 + Read32LittleEndian(rel32);
 498       // To be valid, rel32 target must be within image, and within this
 499       // section.
 500       if (IsValidRVA(target_rva) &&
 501           start_rva <= target_rva && target_rva < end_rva) {
 502         rel32_locations_.push_back(rel32_rva);
 503 #if COURGETTE_HISTOGRAM_TARGETS
 504         ++rel32_target_rvas_[target_rva];
 505 #endif
 506         p = rel32 + 4;
 507         continue;
 508       }
 509     }
 510     p += 1;
 511   }
 512 }
 513
 514 CheckBool DisassemblerWin32X86::ParseNonSectionFileRegion(
 515     uint32 start_file_offset,
 516     uint32 end_file_offset,
 517     AssemblyProgram* program) {
 518   if (incomplete_disassembly_)
 519     return true;
 520
 521   const uint8* start = OffsetToPointer(start_file_offset);
 522   const uint8* end = OffsetToPointer(end_file_offset);
 523
 524   const uint8* p = start;
 525
 526   while (p < end) {
 527     if (!program->EmitByteInstruction(*p))
 528       return false;
 529     ++p;
 530   }
 531
 532   return true;
 533 }
 534
 535 CheckBool DisassemblerWin32X86::ParseFileRegion(
 536     const Section* section,
 537     uint32 start_file_offset, uint32 end_file_offset,
 538     AssemblyProgram* program) {
 539   RVA relocs_start_rva = base_relocation_table().address_;
 540
 541   const uint8* start_pointer = OffsetToPointer(start_file_offset);
 542   const uint8* end_pointer = OffsetToPointer(end_file_offset);
 543
 544   RVA start_rva = FileOffsetToRVA(start_file_offset);
 545   RVA end_rva = start_rva + section->virtual_size;
 546
 547   // Quick way to convert from Pointer to RVA within a single Section is to
 548   // subtract 'pointer_to_rva'.
 549   const uint8* const adjust_pointer_to_rva = start_pointer - start_rva;
 550
 551   std::vector<RVA>::iterator rel32_pos = rel32_locations_.begin();
 552   std::vector<RVA>::iterator abs32_pos = abs32_locations_.begin();
 553
 554   if (!program->EmitOriginInstruction(start_rva))
 555     return false;
 556
 557   const uint8* p = start_pointer;
 558
 559   while (p < end_pointer) {
 560     RVA current_rva = static_cast<RVA>(p - adjust_pointer_to_rva);
 561
 562     // The base relocation table is usually in the .relocs section, but it could
 563     // actually be anywhere.  Make sure we skip it because we will regenerate it
 564     // during assembly.
 565     if (current_rva == relocs_start_rva) {
 566       if (!program->EmitPeRelocsInstruction())
 567         return false;
 568       uint32 relocs_size = base_relocation_table().size_;
 569       if (relocs_size) {
 570         p += relocs_size;
 571         continue;
 572       }
 573     }
 574
 575     while (abs32_pos != abs32_locations_.end() && *abs32_pos < current_rva)
 576       ++abs32_pos;
 577
 578     if (abs32_pos != abs32_locations_.end() && *abs32_pos == current_rva) {
 579       uint32 target_address = Read32LittleEndian(p);
 580       RVA target_rva = target_address - image_base();
 581       // TODO(sra): target could be Label+offset.  It is not clear how to guess
 582       // which it might be.  We assume offset==0.
 583       if (!program->EmitAbs32(program->FindOrMakeAbs32Label(target_rva)))
 584         return false;
 585       p += 4;
 586       continue;
 587     }
 588
 589     while (rel32_pos != rel32_locations_.end() && *rel32_pos < current_rva)
 590       ++rel32_pos;
 591
 592     if (rel32_pos != rel32_locations_.end() && *rel32_pos == current_rva) {
 593       RVA target_rva = current_rva + 4 + Read32LittleEndian(p);
 594       if (!program->EmitRel32(program->FindOrMakeRel32Label(target_rva)))
 595         return false;
 596       p += 4;
 597       continue;
 598     }
 599
 600     if (incomplete_disassembly_) {
 601       if ((abs32_pos == abs32_locations_.end() || end_rva <= *abs32_pos) &&
 602           (rel32_pos == rel32_locations_.end() || end_rva <= *rel32_pos) &&
 603           (end_rva <= relocs_start_rva || current_rva >= relocs_start_rva)) {
 604         // No more relocs in this section, don't bother encoding bytes.
 605         break;
 606       }
 607     }
 608
 609     if (!program->EmitByteInstruction(*p))
 610       return false;
 611     p += 1;
 612   }
 613
 614   return true;
 615 }
 616
 617 #if COURGETTE_HISTOGRAM_TARGETS
 618 // Histogram is printed to std::cout.  It is purely for debugging the algorithm
 619 // and is only enabled manually in 'exploration' builds.  I don't want to add
 620 // command-line configuration for this feature because this code has to be
 621 // small, which means compiled-out.
 622 void DisassemblerWin32X86::HistogramTargets(const char* kind,
 623                                             const std::map<RVA, int>& map) {
 624   int total = 0;
 625   std::map<int, std::vector<RVA> > h;
 626   for (std::map<RVA, int>::const_iterator p = map.begin();
 627        p != map.end();
 628        ++p) {
 629     h[p->second].push_back(p->first);
 630     total += p->second;
 631   }
 632
 633   std::cout << total << " " << kind << " to "
 634             << map.size() << " unique targets" << std::endl;
 635
 636   std::cout << "indegree: #targets-with-indegree (example)" << std::endl;
 637   const int kFirstN = 15;
 638   bool someSkipped = false;
 639   int index = 0;
 640   for (std::map<int, std::vector<RVA> >::reverse_iterator p = h.rbegin();
 641        p != h.rend();
 642        ++p) {
 643     ++index;
 644     if (index <= kFirstN || p->first <= 3) {
 645       if (someSkipped) {
 646         std::cout << "..." << std::endl;
 647       }
 648       size_t count = p->second.size();
 649       std::cout << std::dec << p->first << ": " << count;
 650       if (count <= 2) {
 651         for (size_t i = 0;  i < count;  ++i)
 652           std::cout << "  " << DescribeRVA(p->second[i]);
 653       }
 654       std::cout << std::endl;
 655       someSkipped = false;
 656     } else {
 657       someSkipped = true;
 658     }
 659   }
 660 }
 661 #endif  // COURGETTE_HISTOGRAM_TARGETS
 662
 663
 664 // DescribeRVA is for debugging only.  I would put it under #ifdef DEBUG except
 665 // that during development I'm finding I need to call it when compiled in
 666 // Release mode.  Hence:
 667 // TODO(sra): make this compile only for debug mode.
 668 std::string DisassemblerWin32X86::DescribeRVA(RVA rva) const {
 669   const Section* section = RVAToSection(rva);
 670   std::ostringstream s;
 671   s << std::hex << rva;
 672   if (section) {
 673     s << " (";
 674     s << SectionName(section) << "+"
 675       << std::hex << (rva - section->virtual_address)
 676       << ")";
 677   }
 678   return s.str();
 679 }
 680
 681 const Section* DisassemblerWin32X86::FindNextSection(uint32 fileOffset) const {
 682   const Section* best = 0;
 683   for (int i = 0; i < number_of_sections_; i++) {
 684     const Section* section = &sections_[i];
 685     if (section->size_of_raw_data > 0) {  // i.e. has data in file.
 686       if (fileOffset <= section->file_offset_of_raw_data) {
 687         if (best == 0 ||
 688             section->file_offset_of_raw_data < best->file_offset_of_raw_data) {
 689           best = section;
 690         }
 691       }
 692     }
 693   }
 694   return best;
 695 }
 696
 697 RVA DisassemblerWin32X86::FileOffsetToRVA(uint32 file_offset) const {
 698   for (int i = 0; i < number_of_sections_; i++) {
 699     const Section* section = &sections_[i];
 700     uint32 offset = file_offset - section->file_offset_of_raw_data;
 701     if (offset < section->size_of_raw_data) {
 702       return section->virtual_address + offset;
 703     }
 704   }
 705   return 0;
 706 }
 707
 708 bool DisassemblerWin32X86::ReadDataDirectory(
 709     int index,
 710     ImageDataDirectory* directory) {
 711
 712   if (index < number_of_data_directories_) {
 713     size_t offset = index * 8 + offset_of_data_directories_;
 714     if (offset >= size_of_optional_header_)
 715       return Bad("number of data directories inconsistent");
 716     const uint8* data_directory = optional_header_ + offset;
 717     if (data_directory < start() ||
 718         data_directory + 8 >= end())
 719       return Bad("data directory outside image");
 720     RVA rva = ReadU32(data_directory, 0);
 721     size_t size  = ReadU32(data_directory, 4);
 722     if (size > size_of_image_)
 723       return Bad("data directory size too big");
 724
 725     // TODO(sra): validate RVA.
 726     directory->address_ = rva;
 727     directory->size_ = static_cast<uint32>(size);
 728     return true;
 729   } else {
 730     directory->address_ = 0;
 731     directory->size_ = 0;
 732     return true;
 733   }
 734 }
 735
 736 }  // namespace courgette