courgette/disassembler_win32_x64.cc

   1 // Copyright 2013 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_x64.h"
   6
   7 #include <algorithm>
   8 #include <string>
   9 #include <vector>
  10
  11 #include "base/basictypes.h"
  12 #include "base/logging.h"
  13 #include "base/numerics/safe_conversions.h"
  14
  15 #include "courgette/assembly_program.h"
  16 #include "courgette/courgette.h"
  17 #include "courgette/encoded_program.h"
  18
  19 namespace courgette {
  20
  21 DisassemblerWin32X64::DisassemblerWin32X64(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 DisassemblerWin32X64::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("32 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 DisassemblerWin32X64::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 DisassemblerWin32X64::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       // TODO(sebmarchand): Skip the relocs that live outside of the image. See
 268       //   the version of this function in disassembler_win32_x86.cc.
 269       if (type == 10) {         // IMAGE_REL_BASED_DIR64
 270         relocs->push_back(rva);
 271       } else if (type == 0) {  // IMAGE_REL_BASED_ABSOLUTE
 272         // Ignore, used as padding.
 273       } else {
 274         // Does not occur in Windows x64 executables.
 275         return Bad("unknown type of reloc");
 276       }
 277     }
 278
 279     block += size;
 280   }
 281
 282   std::sort(relocs->begin(), relocs->end());
 283
 284   return true;
 285 }
 286
 287 const Section* DisassemblerWin32X64::RVAToSection(RVA rva) const {
 288   for (int i = 0; i < number_of_sections_; i++) {
 289     const Section* section = &sections_[i];
 290     uint32 offset = rva - section->virtual_address;
 291     if (offset < section->virtual_size) {
 292       return section;
 293     }
 294   }
 295   return NULL;
 296 }
 297
 298 int DisassemblerWin32X64::RVAToFileOffset(RVA rva) const {
 299   const Section* section = RVAToSection(rva);
 300   if (section) {
 301     uint32 offset = rva - section->virtual_address;
 302     if (offset < section->size_of_raw_data) {
 303       return section->file_offset_of_raw_data + offset;
 304     } else {
 305       return kNoOffset;  // In section but not in file (e.g. uninit data).
 306     }
 307   }
 308
 309   // Small RVA values point into the file header in the loaded image.
 310   // RVA 0 is the module load address which Windows uses as the module handle.
 311   // RVA 2 sometimes occurs, I'm not sure what it is, but it would map into the
 312   // DOS header.
 313   if (rva == 0 || rva == 2)
 314     return rva;
 315
 316   NOTREACHED();
 317   return kNoOffset;
 318 }
 319
 320 const uint8* DisassemblerWin32X64::RVAToPointer(RVA rva) const {
 321   int file_offset = RVAToFileOffset(rva);
 322   if (file_offset == kNoOffset)
 323     return NULL;
 324   else
 325     return OffsetToPointer(file_offset);
 326 }
 327
 328 std::string DisassemblerWin32X64::SectionName(const Section* section) {
 329   if (section == NULL)
 330     return "<none>";
 331   char name[9];
 332   memcpy(name, section->name, 8);
 333   name[8] = '\0';  // Ensure termination.
 334   return name;
 335 }
 336
 337 CheckBool DisassemblerWin32X64::ParseFile(AssemblyProgram* program) {
 338   // Walk all the bytes in the file, whether or not in a section.
 339   uint32 file_offset = 0;
 340   while (file_offset < length()) {
 341     const Section* section = FindNextSection(file_offset);
 342     if (section == NULL) {
 343       // No more sections.  There should not be extra stuff following last
 344       // section.
 345       //   ParseNonSectionFileRegion(file_offset, pe_info().length(), program);
 346       break;
 347     }
 348     if (file_offset < section->file_offset_of_raw_data) {
 349       uint32 section_start_offset = section->file_offset_of_raw_data;
 350       if(!ParseNonSectionFileRegion(file_offset, section_start_offset,
 351                                     program))
 352         return false;
 353
 354       file_offset = section_start_offset;
 355     }
 356     uint32 end = file_offset + section->size_of_raw_data;
 357     if (!ParseFileRegion(section, file_offset, end, program))
 358       return false;
 359     file_offset = end;
 360   }
 361
 362 #if COURGETTE_HISTOGRAM_TARGETS
 363   HistogramTargets("abs32 relocs", abs32_target_rvas_);
 364   HistogramTargets("rel32 relocs", rel32_target_rvas_);
 365 #endif
 366
 367   return true;
 368 }
 369
 370 bool DisassemblerWin32X64::ParseAbs32Relocs() {
 371   abs32_locations_.clear();
 372   if (!ParseRelocs(&abs32_locations_))
 373     return false;
 374
 375 #if COURGETTE_HISTOGRAM_TARGETS
 376   for (size_t i = 0;  i < abs32_locations_.size(); ++i) {
 377     RVA rva = abs32_locations_[i];
 378     // The 4 bytes at the relocation are a reference to some address.
 379     uint32 target_address = Read32LittleEndian(RVAToPointer(rva));
 380     ++abs32_target_rvas_[target_address - image_base()];
 381   }
 382 #endif
 383   return true;
 384 }
 385
 386 void DisassemblerWin32X64::ParseRel32RelocsFromSections() {
 387   uint32 file_offset = 0;
 388   while (file_offset < length()) {
 389     const Section* section = FindNextSection(file_offset);
 390     if (section == NULL)
 391       break;
 392     if (file_offset < section->file_offset_of_raw_data)
 393       file_offset = section->file_offset_of_raw_data;
 394     ParseRel32RelocsFromSection(section);
 395     file_offset += section->size_of_raw_data;
 396   }
 397   std::sort(rel32_locations_.begin(), rel32_locations_.end());
 398
 399 #if COURGETTE_HISTOGRAM_TARGETS
 400   VLOG(1) << "abs32_locations_ " << abs32_locations_.size()
 401           << "\nrel32_locations_ " << rel32_locations_.size()
 402           << "\nabs32_target_rvas_ " << abs32_target_rvas_.size()
 403           << "\nrel32_target_rvas_ " << rel32_target_rvas_.size();
 404
 405   int common = 0;
 406   std::map<RVA, int>::iterator abs32_iter = abs32_target_rvas_.begin();
 407   std::map<RVA, int>::iterator rel32_iter = rel32_target_rvas_.begin();
 408   while (abs32_iter != abs32_target_rvas_.end() &&
 409          rel32_iter != rel32_target_rvas_.end()) {
 410     if (abs32_iter->first < rel32_iter->first)
 411       ++abs32_iter;
 412     else if (rel32_iter->first < abs32_iter->first)
 413       ++rel32_iter;
 414     else {
 415       ++common;
 416       ++abs32_iter;
 417       ++rel32_iter;
 418     }
 419   }
 420   VLOG(1) << "common " << common;
 421 #endif
 422 }
 423
 424 void DisassemblerWin32X64::ParseRel32RelocsFromSection(const Section* section) {
 425   // TODO(sra): use characteristic.
 426   bool isCode = strcmp(section->name, ".text") == 0;
 427   if (!isCode)
 428     return;
 429
 430   uint32 start_file_offset = section->file_offset_of_raw_data;
 431   uint32 end_file_offset = start_file_offset + section->size_of_raw_data;
 432   RVA relocs_start_rva = base_relocation_table().address_;
 433
 434   const uint8* start_pointer = OffsetToPointer(start_file_offset);
 435   const uint8* end_pointer = OffsetToPointer(end_file_offset);
 436
 437   RVA start_rva = FileOffsetToRVA(start_file_offset);
 438   RVA end_rva = start_rva + section->virtual_size;
 439
 440   // Quick way to convert from Pointer to RVA within a single Section is to
 441   // subtract 'pointer_to_rva'.
 442   const uint8* const adjust_pointer_to_rva = start_pointer - start_rva;
 443
 444   std::vector<RVA>::iterator abs32_pos = abs32_locations_.begin();
 445
 446   // Find the rel32 relocations.
 447   const uint8* p = start_pointer;
 448   while (p < end_pointer) {
 449     RVA current_rva = static_cast<RVA>(p - adjust_pointer_to_rva);
 450     if (current_rva == relocs_start_rva) {
 451       uint32 relocs_size = base_relocation_table().size_;
 452       if (relocs_size) {
 453         p += relocs_size;
 454         continue;
 455       }
 456     }
 457
 458     //while (abs32_pos != abs32_locations_.end() && *abs32_pos < current_rva)
 459     //  ++abs32_pos;
 460
 461     // Heuristic discovery of rel32 locations in instruction stream: are the
 462     // next few bytes the start of an instruction containing a rel32
 463     // addressing mode?
 464     const uint8* rel32 = NULL;
 465     bool is_rip_relative = false;
 466
 467     if (p + 5 <= end_pointer) {
 468       if (*p == 0xE8 || *p == 0xE9)  // jmp rel32 and call rel32
 469         rel32 = p + 1;
 470     }
 471     if (p + 6 <= end_pointer) {
 472       if (*p == 0x0F && (*(p + 1) & 0xF0) == 0x80) {  // Jcc long form
 473         if (p[1] != 0x8A && p[1] != 0x8B)  // JPE/JPO unlikely
 474           rel32 = p + 2;
 475       } else if (*p == 0xFF && (*(p + 1) == 0x15 || *(p + 1) == 0x25)) {
 476         // rip relative call/jmp
 477         rel32 = p + 2;
 478         is_rip_relative = true;
 479       }
 480     }
 481     if (p + 7 <= end_pointer) {
 482       if ((*p & 0xFB) == 0x48 && *(p + 1) == 0x8D &&
 483           (*(p + 2) & 0xC7) == 0x05) {
 484         // rip relative lea
 485         rel32 = p + 3;
 486         is_rip_relative = true;
 487       } else if ((*p & 0xFB) == 0x48 && *(p + 1) == 0x8B &&
 488                  (*(p + 2) & 0xC7) == 0x05) {
 489         // rip relative mov
 490         rel32 = p + 3;
 491         is_rip_relative = true;
 492       }
 493     }
 494
 495     if (rel32) {
 496       RVA rel32_rva = static_cast<RVA>(rel32 - adjust_pointer_to_rva);
 497
 498       // Is there an abs32 reloc overlapping the candidate?
 499       while (abs32_pos != abs32_locations_.end() && *abs32_pos < rel32_rva - 3)
 500         ++abs32_pos;
 501       // Now: (*abs32_pos > rel32_rva - 4) i.e. the lowest addressed 4-byte
 502       // region that could overlap rel32_rva.
 503       if (abs32_pos != abs32_locations_.end()) {
 504         if (*abs32_pos < rel32_rva + 4) {
 505           // Beginning of abs32 reloc is before end of rel32 reloc so they
 506           // overlap.  Skip four bytes past the abs32 reloc.
 507           p += (*abs32_pos + 4) - current_rva;
 508           continue;
 509         }
 510       }
 511
 512       RVA target_rva = rel32_rva + 4 + Read32LittleEndian(rel32);
 513       // To be valid, rel32 target must be within image, and within this
 514       // section.
 515       if (IsValidRVA(target_rva) &&
 516           (is_rip_relative ||
 517            (start_rva <= target_rva && target_rva < end_rva))) {
 518         rel32_locations_.push_back(rel32_rva);
 519 #if COURGETTE_HISTOGRAM_TARGETS
 520         ++rel32_target_rvas_[target_rva];
 521 #endif
 522         p = rel32 + 4;
 523         continue;
 524       }
 525     }
 526     p += 1;
 527   }
 528 }
 529
 530 CheckBool DisassemblerWin32X64::ParseNonSectionFileRegion(
 531     uint32 start_file_offset,
 532     uint32 end_file_offset,
 533     AssemblyProgram* program) {
 534   if (incomplete_disassembly_)
 535     return true;
 536
 537   if (end_file_offset > start_file_offset) {
 538     if (!program->EmitBytesInstruction(OffsetToPointer(start_file_offset),
 539                                        end_file_offset - start_file_offset)) {
 540       return false;
 541     }
 542   }
 543
 544   return true;
 545 }
 546
 547 CheckBool DisassemblerWin32X64::ParseFileRegion(
 548     const Section* section,
 549     uint32 start_file_offset, uint32 end_file_offset,
 550     AssemblyProgram* program) {
 551   RVA relocs_start_rva = base_relocation_table().address_;
 552
 553   const uint8* start_pointer = OffsetToPointer(start_file_offset);
 554   const uint8* end_pointer = OffsetToPointer(end_file_offset);
 555
 556   RVA start_rva = FileOffsetToRVA(start_file_offset);
 557   RVA end_rva = start_rva + section->virtual_size;
 558
 559   // Quick way to convert from Pointer to RVA within a single Section is to
 560   // subtract 'pointer_to_rva'.
 561   const uint8* const adjust_pointer_to_rva = start_pointer - start_rva;
 562
 563   std::vector<RVA>::iterator rel32_pos = rel32_locations_.begin();
 564   std::vector<RVA>::iterator abs32_pos = abs32_locations_.begin();
 565
 566   if (!program->EmitOriginInstruction(start_rva))
 567     return false;
 568
 569   const uint8* p = start_pointer;
 570
 571   while (p < end_pointer) {
 572     RVA current_rva = static_cast<RVA>(p - adjust_pointer_to_rva);
 573
 574     // The base relocation table is usually in the .relocs section, but it could
 575     // actually be anywhere.  Make sure we skip it because we will regenerate it
 576     // during assembly.
 577     if (current_rva == relocs_start_rva) {
 578       if (!program->EmitPeRelocsInstruction())
 579         return false;
 580       uint32 relocs_size = base_relocation_table().size_;
 581       if (relocs_size) {
 582         p += relocs_size;
 583         continue;
 584       }
 585     }
 586
 587     while (abs32_pos != abs32_locations_.end() && *abs32_pos < current_rva)
 588       ++abs32_pos;
 589
 590     if (abs32_pos != abs32_locations_.end() && *abs32_pos == current_rva) {
 591       uint64 target_address = Read64LittleEndian(p);
 592       RVA target_rva = base::checked_cast<RVA>(target_address - image_base());
 593       // TODO(sra): target could be Label+offset.  It is not clear how to guess
 594       // which it might be.  We assume offset==0.
 595       if (!program->EmitAbs64(program->FindOrMakeAbs32Label(target_rva)))
 596         return false;
 597       p += 8;
 598       continue;
 599     }
 600
 601     while (rel32_pos != rel32_locations_.end() && *rel32_pos < current_rva)
 602       ++rel32_pos;
 603
 604     if (rel32_pos != rel32_locations_.end() && *rel32_pos == current_rva) {
 605       RVA target_rva = current_rva + 4 + Read32LittleEndian(p);
 606       if (!program->EmitRel32(program->FindOrMakeRel32Label(target_rva)))
 607         return false;
 608       p += 4;
 609       continue;
 610     }
 611
 612     if (incomplete_disassembly_) {
 613       if ((abs32_pos == abs32_locations_.end() || end_rva <= *abs32_pos) &&
 614           (rel32_pos == rel32_locations_.end() || end_rva <= *rel32_pos) &&
 615           (end_rva <= relocs_start_rva || current_rva >= relocs_start_rva)) {
 616         // No more relocs in this section, don't bother encoding bytes.
 617         break;
 618       }
 619     }
 620
 621     if (!program->EmitByteInstruction(*p))
 622       return false;
 623     p += 1;
 624   }
 625
 626   return true;
 627 }
 628
 629 #if COURGETTE_HISTOGRAM_TARGETS
 630 // Histogram is printed to std::cout.  It is purely for debugging the algorithm
 631 // and is only enabled manually in 'exploration' builds.  I don't want to add
 632 // command-line configuration for this feature because this code has to be
 633 // small, which means compiled-out.
 634 void DisassemblerWin32X64::HistogramTargets(const char* kind,
 635                                             const std::map<RVA, int>& map) {
 636   int total = 0;
 637   std::map<int, std::vector<RVA> > h;
 638   for (std::map<RVA, int>::const_iterator p = map.begin();
 639        p != map.end();
 640        ++p) {
 641     h[p->second].push_back(p->first);
 642     total += p->second;
 643   }
 644
 645   std::cout << total << " " << kind << " to "
 646             << map.size() << " unique targets" << std::endl;
 647
 648   std::cout << "indegree: #targets-with-indegree (example)" << std::endl;
 649   const int kFirstN = 15;
 650   bool someSkipped = false;
 651   int index = 0;
 652   for (std::map<int, std::vector<RVA> >::reverse_iterator p = h.rbegin();
 653        p != h.rend();
 654        ++p) {
 655     ++index;
 656     if (index <= kFirstN || p->first <= 3) {
 657       if (someSkipped) {
 658         std::cout << "..." << std::endl;
 659       }
 660       size_t count = p->second.size();
 661       std::cout << std::dec << p->first << ": " << count;
 662       if (count <= 2) {
 663         for (size_t i = 0;  i < count;  ++i)
 664           std::cout << "  " << DescribeRVA(p->second[i]);
 665       }
 666       std::cout << std::endl;
 667       someSkipped = false;
 668     } else {
 669       someSkipped = true;
 670     }
 671   }
 672 }
 673 #endif  // COURGETTE_HISTOGRAM_TARGETS
 674
 675
 676 // DescribeRVA is for debugging only.  I would put it under #ifdef DEBUG except
 677 // that during development I'm finding I need to call it when compiled in
 678 // Release mode.  Hence:
 679 // TODO(sra): make this compile only for debug mode.
 680 std::string DisassemblerWin32X64::DescribeRVA(RVA rva) const {
 681   const Section* section = RVAToSection(rva);
 682   std::ostringstream s;
 683   s << std::hex << rva;
 684   if (section) {
 685     s << " (";
 686     s << SectionName(section) << "+"
 687       << std::hex << (rva - section->virtual_address)
 688       << ")";
 689   }
 690   return s.str();
 691 }
 692
 693 const Section* DisassemblerWin32X64::FindNextSection(uint32 fileOffset) const {
 694   const Section* best = 0;
 695   for (int i = 0; i < number_of_sections_; i++) {
 696     const Section* section = &sections_[i];
 697     if (section->size_of_raw_data > 0) {  // i.e. has data in file.
 698       if (fileOffset <= section->file_offset_of_raw_data) {
 699         if (best == 0 ||
 700             section->file_offset_of_raw_data < best->file_offset_of_raw_data) {
 701           best = section;
 702         }
 703       }
 704     }
 705   }
 706   return best;
 707 }
 708
 709 RVA DisassemblerWin32X64::FileOffsetToRVA(uint32 file_offset) const {
 710   for (int i = 0; i < number_of_sections_; i++) {
 711     const Section* section = &sections_[i];
 712     uint32 offset = file_offset - section->file_offset_of_raw_data;
 713     if (offset < section->size_of_raw_data) {
 714       return section->virtual_address + offset;
 715     }
 716   }
 717   return 0;
 718 }
 719
 720 bool DisassemblerWin32X64::ReadDataDirectory(
 721     int index,
 722     ImageDataDirectory* directory) {
 723
 724   if (index < number_of_data_directories_) {
 725     size_t offset = index * 8 + offset_of_data_directories_;
 726     if (offset >= size_of_optional_header_)
 727       return Bad("number of data directories inconsistent");
 728     const uint8* data_directory = optional_header_ + offset;
 729     if (data_directory < start() ||
 730         data_directory + 8 >= end())
 731       return Bad("data directory outside image");
 732     RVA rva = ReadU32(data_directory, 0);
 733     size_t size  = ReadU32(data_directory, 4);
 734     if (size > size_of_image_)
 735       return Bad("data directory size too big");
 736
 737     // TODO(sra): validate RVA.
 738     directory->address_ = rva;
 739     directory->size_ = static_cast<uint32>(size);
 740     return true;
 741   } else {
 742     directory->address_ = 0;
 743     directory->size_ = 0;
 744     return true;
 745   }
 746 }
 747
 748 }  // namespace courgette