courgette/disassembler_elf_32.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_elf_32.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/memory/scoped_vector.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 DisassemblerElf32::DisassemblerElf32(const void* start, size_t length)
  22   : Disassembler(start, length),
  23     header_(NULL),
  24     section_header_table_(NULL),
  25     section_header_table_size_(0),
  26     program_header_table_(NULL),
  27     program_header_table_size_(0),
  28     default_string_section_(NULL) {
  29 }
  30
  31 bool DisassemblerElf32::ParseHeader() {
  32   if (length() < sizeof(Elf32_Ehdr))
  33     return Bad("Too small");
  34
  35   header_ = (Elf32_Ehdr *)start();
  36
  37   // Have magic for elf header?
  38   if (header_->e_ident[0] != 0x7f ||
  39       header_->e_ident[1] != 'E' ||
  40       header_->e_ident[2] != 'L' ||
  41       header_->e_ident[3] != 'F')
  42     return Bad("No Magic Number");
  43
  44   if (header_->e_type != ET_EXEC &&
  45       header_->e_type != ET_DYN)
  46     return Bad("Not an executable file or shared library");
  47
  48   if (header_->e_machine != ElfEM())
  49     return Bad("Not a supported architecture");
  50
  51   if (header_->e_version != 1)
  52     return Bad("Unknown file version");
  53
  54   if (header_->e_shentsize != sizeof(Elf32_Shdr))
  55     return Bad("Unexpected section header size");
  56
  57   if (header_->e_shoff >= length())
  58     return Bad("Out of bounds section header table offset");
  59
  60   section_header_table_ = (Elf32_Shdr *)OffsetToPointer(header_->e_shoff);
  61   section_header_table_size_ = header_->e_shnum;
  62
  63   if ((header_->e_shoff + header_->e_shnum ) >= length())
  64     return Bad("Out of bounds section header table");
  65
  66   if (header_->e_phoff >= length())
  67     return Bad("Out of bounds program header table offset");
  68
  69   program_header_table_ = (Elf32_Phdr *)OffsetToPointer(header_->e_phoff);
  70   program_header_table_size_ = header_->e_phnum;
  71
  72   if ((header_->e_phoff + header_->e_phnum) >= length())
  73     return Bad("Out of bounds program header table");
  74
  75   default_string_section_ = (const char *)SectionBody((int)header_->e_shstrndx);
  76
  77   ReduceLength(DiscoverLength());
  78
  79   return Good();
  80 }
  81
  82 bool DisassemblerElf32::Disassemble(AssemblyProgram* target) {
  83   if (!ok())
  84     return false;
  85
  86   // The Image Base is always 0 for ELF Executables
  87   target->set_image_base(0);
  88
  89   if (!ParseAbs32Relocs())
  90     return false;
  91
  92   if (!ParseRel32RelocsFromSections())
  93     return false;
  94
  95   if (!ParseFile(target))
  96     return false;
  97
  98   target->DefaultAssignIndexes();
  99
 100   return true;
 101 }
 102
 103 uint32 DisassemblerElf32::DiscoverLength() {
 104   uint32 result = 0;
 105
 106   // Find the end of the last section
 107   for (int section_id = 0; section_id < SectionHeaderCount(); section_id++) {
 108     const Elf32_Shdr *section_header = SectionHeader(section_id);
 109
 110     if (section_header->sh_type == SHT_NOBITS)
 111       continue;
 112
 113     uint32 section_end = section_header->sh_offset + section_header->sh_size;
 114
 115     if (section_end > result)
 116       result = section_end;
 117   }
 118
 119   // Find the end of the last segment
 120   for (int i = 0; i < ProgramSegmentHeaderCount(); i++) {
 121     const Elf32_Phdr *segment_header = ProgramSegmentHeader(i);
 122
 123     uint32 segment_end = segment_header->p_offset + segment_header->p_filesz;
 124
 125     if (segment_end > result)
 126       result = segment_end;
 127   }
 128
 129   uint32 section_table_end = header_->e_shoff +
 130                              (header_->e_shnum * sizeof(Elf32_Shdr));
 131   if (section_table_end > result)
 132     result = section_table_end;
 133
 134   uint32 segment_table_end = header_->e_phoff +
 135                              (header_->e_phnum * sizeof(Elf32_Phdr));
 136   if (segment_table_end > result)
 137     result = segment_table_end;
 138
 139   return result;
 140 }
 141
 142 CheckBool DisassemblerElf32::IsValidRVA(RVA rva) const {
 143
 144   // It's valid if it's contained in any program segment
 145   for (int i = 0; i < ProgramSegmentHeaderCount(); i++) {
 146     const Elf32_Phdr *segment_header = ProgramSegmentHeader(i);
 147
 148     if (segment_header->p_type != PT_LOAD)
 149       continue;
 150
 151     Elf32_Addr begin = segment_header->p_vaddr;
 152     Elf32_Addr end = segment_header->p_vaddr + segment_header->p_memsz;
 153
 154     if (rva >= begin && rva < end)
 155       return true;
 156   }
 157
 158   return false;
 159 }
 160
 161 // Returns RVA for an in memory address, or NULL.
 162 CheckBool DisassemblerElf32::RVAToFileOffset(Elf32_Addr addr,
 163                                                 size_t* result) const {
 164
 165   for (int i = 0; i < ProgramSegmentHeaderCount(); i++) {
 166     Elf32_Addr begin = ProgramSegmentMemoryBegin(i);
 167     Elf32_Addr end = begin + ProgramSegmentMemorySize(i);
 168
 169     if (addr >= begin  && addr < end) {
 170       Elf32_Addr offset = addr - begin;
 171
 172       if (offset < ProgramSegmentFileSize(i)) {
 173         *result = ProgramSegmentFileOffset(i) + offset;
 174         return true;
 175       }
 176     }
 177   }
 178
 179   return false;
 180 }
 181
 182 RVA DisassemblerElf32::FileOffsetToRVA(size_t offset) const {
 183   // File offsets can be 64 bit values, but we are dealing with 32
 184   // bit executables and so only need to support 32bit file sizes.
 185   uint32 offset32 = (uint32)offset;
 186
 187   for (int i = 0; i < SectionHeaderCount(); i++) {
 188
 189     const Elf32_Shdr *section_header = SectionHeader(i);
 190
 191     // These can appear to have a size in the file, but don't.
 192     if (section_header->sh_type == SHT_NOBITS)
 193       continue;
 194
 195     Elf32_Off section_begin = section_header->sh_offset;
 196     Elf32_Off section_end = section_begin + section_header->sh_size;
 197
 198     if (offset32 >= section_begin && offset32 < section_end) {
 199       return section_header->sh_addr + (offset32 - section_begin);
 200     }
 201   }
 202
 203   return 0;
 204 }
 205
 206 CheckBool DisassemblerElf32::RVAsToOffsets(std::vector<RVA>* rvas,
 207                                            std::vector<size_t>* offsets) {
 208   offsets->clear();
 209
 210   for (std::vector<RVA>::iterator rva = rvas->begin();
 211        rva != rvas->end();
 212        rva++) {
 213
 214     size_t offset;
 215
 216     if (!RVAToFileOffset(*rva, &offset))
 217       return false;
 218
 219     offsets->push_back(offset);
 220   }
 221
 222   return true;
 223 }
 224
 225 CheckBool DisassemblerElf32::RVAsToOffsets(ScopedVector<TypedRVA>* rvas) {
 226   for (ScopedVector<TypedRVA>::iterator rva = rvas->begin();
 227        rva != rvas->end();
 228        rva++) {
 229
 230     size_t offset;
 231
 232     if (!RVAToFileOffset((*rva)->rva(), &offset))
 233       return false;
 234
 235     (*rva)->set_offset(offset);
 236   }
 237
 238   return true;
 239 }
 240
 241 CheckBool DisassemblerElf32::ParseFile(AssemblyProgram* program) {
 242   // Walk all the bytes in the file, whether or not in a section.
 243   uint32 file_offset = 0;
 244
 245   std::vector<size_t> abs_offsets;
 246
 247   if (!RVAsToOffsets(&abs32_locations_, &abs_offsets))
 248     return false;
 249
 250   if (!RVAsToOffsets(&rel32_locations_))
 251     return false;
 252
 253   std::vector<size_t>::iterator current_abs_offset = abs_offsets.begin();
 254   ScopedVector<TypedRVA>::iterator current_rel = rel32_locations_.begin();
 255
 256   std::vector<size_t>::iterator end_abs_offset = abs_offsets.end();
 257   ScopedVector<TypedRVA>::iterator end_rel = rel32_locations_.end();
 258
 259   for (int section_id = 0;
 260        section_id < SectionHeaderCount();
 261        section_id++) {
 262
 263     const Elf32_Shdr *section_header = SectionHeader(section_id);
 264
 265     if (!ParseSimpleRegion(file_offset,
 266                            section_header->sh_offset,
 267                            program))
 268       return false;
 269     file_offset = section_header->sh_offset;
 270
 271     switch (section_header->sh_type) {
 272       case SHT_REL:
 273         if (!ParseRelocationSection(section_header, program))
 274           return false;
 275         file_offset = section_header->sh_offset + section_header->sh_size;
 276         break;
 277       case SHT_PROGBITS:
 278         if (!ParseProgbitsSection(section_header,
 279                                   &current_abs_offset, end_abs_offset,
 280                                   &current_rel, end_rel,
 281                                   program))
 282           return false;
 283         file_offset = section_header->sh_offset + section_header->sh_size;
 284         break;
 285       case SHT_NOBITS:
 286         // Fall through
 287       case SHT_INIT_ARRAY:
 288         // Fall through
 289       case SHT_FINI_ARRAY:
 290         while (current_abs_offset != end_abs_offset &&
 291                *current_abs_offset >= section_header->sh_offset &&
 292                *current_abs_offset <
 293                (section_header->sh_offset + section_header->sh_size)) {
 294           // Skip any abs_offsets appear in the unsupported INIT_ARRAY section
 295           VLOG(1) << "Skipping relocation entry for unsupported section: " <<
 296             section_header->sh_type;
 297           current_abs_offset++;
 298         }
 299         break;
 300       default:
 301         if (current_abs_offset != end_abs_offset &&
 302                *current_abs_offset >= section_header->sh_offset &&
 303                *current_abs_offset <
 304                (section_header->sh_offset + section_header->sh_size))
 305           VLOG(1) << "Relocation address in unrecognized ELF section: " << \
 306             section_header->sh_type;
 307       break;
 308     }
 309   }
 310
 311   // Rest of the file past the last section
 312   if (!ParseSimpleRegion(file_offset,
 313                          length(),
 314                          program))
 315     return false;
 316
 317   // Make certain we consume all of the relocations as expected
 318   return (current_abs_offset == end_abs_offset);
 319 }
 320
 321 CheckBool DisassemblerElf32::ParseProgbitsSection(
 322     const Elf32_Shdr *section_header,
 323     std::vector<size_t>::iterator* current_abs_offset,
 324     std::vector<size_t>::iterator end_abs_offset,
 325     ScopedVector<TypedRVA>::iterator* current_rel,
 326     ScopedVector<TypedRVA>::iterator end_rel,
 327     AssemblyProgram* program) {
 328
 329   // Walk all the bytes in the file, whether or not in a section.
 330   size_t file_offset = section_header->sh_offset;
 331   size_t section_end = section_header->sh_offset + section_header->sh_size;
 332
 333   Elf32_Addr origin = section_header->sh_addr;
 334   size_t origin_offset = section_header->sh_offset;
 335   if (!program->EmitOriginInstruction(origin))
 336     return false;
 337
 338   while (file_offset < section_end) {
 339
 340     if (*current_abs_offset != end_abs_offset &&
 341         file_offset > **current_abs_offset)
 342       return false;
 343
 344     while (*current_rel != end_rel &&
 345            file_offset > (**current_rel)->get_offset()) {
 346       (*current_rel)++;
 347     }
 348
 349     size_t next_relocation = section_end;
 350
 351     if (*current_abs_offset != end_abs_offset &&
 352         next_relocation > **current_abs_offset)
 353       next_relocation = **current_abs_offset;
 354
 355     // Rel offsets are heuristically derived, and might (incorrectly) overlap
 356     // an Abs value, or the end of the section, so +3 to make sure there is
 357     // room for the full 4 byte value.
 358     if (*current_rel != end_rel &&
 359         next_relocation > ((**current_rel)->get_offset() + 3))
 360       next_relocation = (**current_rel)->get_offset();
 361
 362     if (next_relocation > file_offset) {
 363       if (!ParseSimpleRegion(file_offset, next_relocation, program))
 364         return false;
 365
 366       file_offset = next_relocation;
 367       continue;
 368     }
 369
 370     if (*current_abs_offset != end_abs_offset &&
 371         file_offset == **current_abs_offset) {
 372
 373       const uint8* p = OffsetToPointer(file_offset);
 374       RVA target_rva = Read32LittleEndian(p);
 375
 376       if (!program->EmitAbs32(program->FindOrMakeAbs32Label(target_rva)))
 377         return false;
 378       file_offset += sizeof(RVA);
 379       (*current_abs_offset)++;
 380       continue;
 381     }
 382
 383     if (*current_rel != end_rel &&
 384         file_offset == (**current_rel)->get_offset()) {
 385
 386       uint32 relative_target = (**current_rel)->relative_target();
 387       // This cast is for 64 bit systems, and is only safe because we
 388       // are working on 32 bit executables.
 389       RVA target_rva = (RVA)(origin + (file_offset - origin_offset) +
 390                              relative_target);
 391
 392       if (! (**current_rel)->EmitInstruction(program, target_rva))
 393         return false;
 394       file_offset += (**current_rel)->op_size();
 395       (*current_rel)++;
 396       continue;
 397     }
 398   }
 399
 400   // Rest of the section (if any)
 401   return ParseSimpleRegion(file_offset, section_end, program);
 402 }
 403
 404 CheckBool DisassemblerElf32::ParseSimpleRegion(
 405     size_t start_file_offset,
 406     size_t end_file_offset,
 407     AssemblyProgram* program) {
 408
 409   const uint8* start = OffsetToPointer(start_file_offset);
 410   const uint8* end = OffsetToPointer(end_file_offset);
 411
 412   // Callers don't guarantee start < end
 413   if (start >= end) return true;
 414
 415   const ptrdiff_t len = end - start;  // Works because vars are byte pointers
 416
 417   if (!program->EmitBytesInstruction(start, len))
 418     return false;
 419
 420   return true;
 421 }
 422
 423 CheckBool DisassemblerElf32::ParseAbs32Relocs() {
 424   abs32_locations_.clear();
 425
 426   // Loop through sections for relocation sections
 427   for (int section_id = 0; section_id < SectionHeaderCount(); section_id++) {
 428     const Elf32_Shdr *section_header = SectionHeader(section_id);
 429
 430     if (section_header->sh_type == SHT_REL) {
 431
 432       Elf32_Rel *relocs_table = (Elf32_Rel *)SectionBody(section_id);
 433
 434       int relocs_table_count = section_header->sh_size /
 435                                section_header->sh_entsize;
 436
 437       // Elf32_Word relocation_section_id = section_header->sh_info;
 438
 439       // Loop through relocation objects in the relocation section
 440       for (int rel_id = 0; rel_id < relocs_table_count; rel_id++) {
 441         RVA rva;
 442
 443         // Quite a few of these conversions fail, and we simply skip
 444         // them, that's okay.
 445         if (RelToRVA(relocs_table[rel_id], &rva) && CheckSection(rva))
 446           abs32_locations_.push_back(rva);
 447       }
 448     }
 449   }
 450
 451   std::sort(abs32_locations_.begin(), abs32_locations_.end());
 452   return true;
 453 }
 454
 455 CheckBool DisassemblerElf32::CheckSection(RVA rva) {
 456   size_t offset;
 457
 458   if (!RVAToFileOffset(rva, &offset)) {
 459     return false;
 460   }
 461
 462   for (int section_id = 0;
 463        section_id < SectionHeaderCount();
 464        section_id++) {
 465
 466     const Elf32_Shdr *section_header = SectionHeader(section_id);
 467
 468     if (offset >= section_header->sh_offset &&
 469         offset < (section_header->sh_offset + section_header->sh_size)) {
 470       switch (section_header->sh_type) {
 471         case SHT_REL:
 472           // Fall-through
 473         case SHT_PROGBITS:
 474           return true;
 475       }
 476     }
 477   }
 478
 479   return false;
 480 }
 481
 482 CheckBool DisassemblerElf32::ParseRel32RelocsFromSections() {
 483
 484   rel32_locations_.clear();
 485
 486   // Loop through sections for relocation sections
 487   for (int section_id = 0;
 488        section_id < SectionHeaderCount();
 489        section_id++) {
 490
 491     const Elf32_Shdr *section_header = SectionHeader(section_id);
 492
 493     if (section_header->sh_type != SHT_PROGBITS)
 494       continue;
 495
 496     if (!ParseRel32RelocsFromSection(section_header))
 497       return false;
 498   }
 499
 500   std::sort(rel32_locations_.begin(),
 501             rel32_locations_.end(),
 502             TypedRVA::IsLessThan);
 503   return true;
 504 }
 505
 506 }  // namespace courgette