DXR is a code search and navigation tool aimed at making sense of large projects. It supports full-text and regex searches as well as structural queries.

Mercurial (4a108e94d3e2)

VCS Links

Line Code
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038
/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* vim: set ts=8 sts=2 et sw=2 tw=80: */

// Copyright (c) 2006, 2011, 2012 Google Inc.
// All rights reserved.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
//     * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//     * Redistributions in binary form must reproduce the above
// copyright notice, this list of conditions and the following disclaimer
// in the documentation and/or other materials provided with the
// distribution.
//     * Neither the name of Google Inc. nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

// Restructured in 2009 by: Jim Blandy <jimb@mozilla.com> <jimb@red-bean.com>

// (derived from)
// dump_symbols.cc: implement google_breakpad::WriteSymbolFile:
// Find all the debugging info in a file and dump it as a Breakpad symbol file.
//
// dump_symbols.h: Read debugging information from an ELF file, and write
// it out as a Breakpad symbol file.

// This file is derived from the following files in
// toolkit/crashreporter/google-breakpad:
//   src/common/linux/dump_symbols.cc
//   src/common/linux/elfutils.cc
//   src/common/linux/file_id.cc

#include <errno.h>
#include <fcntl.h>
#include <stdio.h>
#include <string.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <unistd.h>
#include <arpa/inet.h>

#include <set>
#include <string>
#include <vector>

#include "mozilla/Assertions.h"

#include "LulPlatformMacros.h"
#include "LulCommonExt.h"
#include "LulDwarfExt.h"
#if defined(LUL_PLAT_arm_android)
# include "LulExidxExt.h"
#endif
#include "LulElfInt.h"
#include "LulMainInt.h"


#if defined(LUL_PLAT_arm_android) && !defined(SHT_ARM_EXIDX)
// bionic and older glibsc don't define it
# define SHT_ARM_EXIDX (SHT_LOPROC + 1)
#endif


// This namespace contains helper functions.
namespace {

using lul::DwarfCFIToModule;
using lul::FindElfSectionByName;
using lul::GetOffset;
using lul::IsValidElf;
using lul::Module;
using lul::UniqueString;
using lul::scoped_ptr;
using lul::Summariser;
using std::string;
using std::vector;
using std::set;

//
// FDWrapper
//
// Wrapper class to make sure opened file is closed.
//
class FDWrapper {
 public:
  explicit FDWrapper(int fd) :
    fd_(fd) {}
  ~FDWrapper() {
    if (fd_ != -1)
      close(fd_);
  }
  int get() {
    return fd_;
  }
  int release() {
    int fd = fd_;
    fd_ = -1;
    return fd;
  }
 private:
  int fd_;
};

//
// MmapWrapper
//
// Wrapper class to make sure mapped regions are unmapped.
//
class MmapWrapper {
 public:
  MmapWrapper() : is_set_(false) {}
  ~MmapWrapper() {
    if (is_set_ && base_ != NULL) {
      MOZ_ASSERT(size_ > 0);
      munmap(base_, size_);
    }
  }
  void set(void *mapped_address, size_t mapped_size) {
    is_set_ = true;
    base_ = mapped_address;
    size_ = mapped_size;
  }
  void release() {
    MOZ_ASSERT(is_set_);
    is_set_ = false;
    base_ = NULL;
    size_ = 0;
  }

 private:
  bool is_set_;
  void *base_;
  size_t size_;
};


// Set NUM_DW_REGNAMES to be the number of Dwarf register names
// appropriate to the machine architecture given in HEADER.  Return
// true on success, or false if HEADER's machine architecture is not
// supported.
template<typename ElfClass>
bool DwarfCFIRegisterNames(const typename ElfClass::Ehdr* elf_header,
                           unsigned int* num_dw_regnames) {
  switch (elf_header->e_machine) {
    case EM_386:
      *num_dw_regnames = DwarfCFIToModule::RegisterNames::I386();
      return true;
    case EM_ARM:
      *num_dw_regnames = DwarfCFIToModule::RegisterNames::ARM();
      return true;
    case EM_X86_64:
      *num_dw_regnames = DwarfCFIToModule::RegisterNames::X86_64();
      return true;
    default:
      MOZ_ASSERT(0);
      return false;
  }
}

template<typename ElfClass>
bool LoadDwarfCFI(const string& dwarf_filename,
                  const typename ElfClass::Ehdr* elf_header,
                  const char* section_name,
                  const typename ElfClass::Shdr* section,
                  const bool eh_frame,
                  const typename ElfClass::Shdr* got_section,
                  const typename ElfClass::Shdr* text_section,
                  const bool big_endian,
                  SecMap* smap,
                  uintptr_t text_bias,
                  void (*log)(const char*)) {
  // Find the appropriate set of register names for this file's
  // architecture.
  unsigned int num_dw_regs = 0;
  if (!DwarfCFIRegisterNames<ElfClass>(elf_header, &num_dw_regs)) {
    fprintf(stderr, "%s: unrecognized ELF machine architecture '%d';"
            " cannot convert DWARF call frame information\n",
            dwarf_filename.c_str(), elf_header->e_machine);
    return false;
  }

  const lul::Endianness endianness
    = big_endian ? lul::ENDIANNESS_BIG : lul::ENDIANNESS_LITTLE;

  // Find the call frame information and its size.
  const char* cfi =
      GetOffset<ElfClass, char>(elf_header, section->sh_offset);
  size_t cfi_size = section->sh_size;

  // Plug together the parser, handler, and their entourages.

  // Here's a summariser, which will receive the output of the
  // parser, create summaries, and add them to |smap|.
  Summariser* summ = new Summariser(smap, text_bias, log);

  DwarfCFIToModule::Reporter module_reporter(log, dwarf_filename, section_name);
  DwarfCFIToModule handler(num_dw_regs, &module_reporter, summ);
  lul::ByteReader byte_reader(endianness);

  byte_reader.SetAddressSize(ElfClass::kAddrSize);

  // Provide the base addresses for .eh_frame encoded pointers, if
  // possible.
  byte_reader.SetCFIDataBase(section->sh_addr, cfi);
  if (got_section)
    byte_reader.SetDataBase(got_section->sh_addr);
  if (text_section)
    byte_reader.SetTextBase(text_section->sh_addr);

  lul::CallFrameInfo::Reporter dwarf_reporter(log, dwarf_filename,
                                              section_name);
  lul::CallFrameInfo parser(cfi, cfi_size,
                            &byte_reader, &handler, &dwarf_reporter,
                            eh_frame);
  parser.Start();

  delete summ;
  return true;
}

#if defined(LUL_PLAT_arm_android)
template<typename ElfClass>
bool LoadARMexidx(const typename ElfClass::Ehdr* elf_header,
                  const typename ElfClass::Shdr* exidx_section,
                  const typename ElfClass::Shdr* extab_section,
                  uintptr_t text_bias,
                  uintptr_t rx_avma, size_t rx_size,
                  SecMap* smap,
                  void (*log)(const char*)) {
  // To do this properly we need to know:
  // * the bounds of the .ARM.exidx section in the process image
  // * the bounds of the .ARM.extab section in the process image
  // * the vma of the last byte in the text section associated with the .exidx
  // The first two are easy.  The third is a bit tricky.  If we can't
  // figure out what it is, just pass in zero.
  // Note that we are reading EXIDX directly out of the mapped in
  // executable image.  Unlike with the CFI reader, there is no
  // auxiliary, temporary mapping used to read the unwind data.
  //
  // An .exidx section is always required, but the .extab section
  // can be optionally omitted, provided that .exidx does not refer
  // to it.  If the .exidx is erroneous and does refer to .extab even
  // though .extab is missing, the range checks done by GET_EX_U32 in
  // ExceptionTableInfo::ExtabEntryExtract should prevent any invalid
  // memory accesses, and cause the .extab to be rejected as invalid.

  uintptr_t exidx_svma = exidx_section->sh_addr;
  uintptr_t exidx_avma = exidx_svma + text_bias;
  size_t    exidx_size = exidx_section->sh_size;

  uintptr_t extab_svma = 0;
  uintptr_t extab_avma = 0;
  size_t    extab_size = 0;
  if (extab_section) {
    extab_svma = extab_section->sh_addr;
    extab_avma = extab_svma + text_bias;
    extab_size = extab_section->sh_size;
  }

  // Because we are reading EXIDX directly out of the executing image,
  // we need to be careful to check that the relevant sections have
  // really been mapped with r permissions, so as to guarantee that
  // reading them won't segfault.  Do this by checking that rx mapped
  // area covers the exidx and extab as mapped in.

  if (rx_size == 0)
    // This seems sufficiently bogus that we shouldn't proceed further.
    return false;

  if (exidx_size == 0)
    // There's no EXIDX data.  No point in continuing.
    return false;

  if (!(exidx_avma >= rx_avma && exidx_avma + exidx_size <= rx_avma + rx_size))
    // The mapped .exidx isn't entirely inside the rx area.
    return false;

  if (extab_section &&
      !(extab_avma >= rx_avma && extab_avma + extab_size <= rx_avma + rx_size))
    // There an .extab section, but it isn't entirely inside the rx area.
    return false;

  // The sh_link field of the exidx section gives the section number
  // for the associated text section.
  uint32_t exidx_text_last_avma = 0;
  int exidx_text_sno = exidx_section->sh_link;
  typedef typename ElfClass::Shdr Shdr;
  // |sections| points to the section header table
  const Shdr* sections
    = GetOffset<ElfClass, Shdr>(elf_header, elf_header->e_shoff);
  const int num_sections = elf_header->e_shnum;
  if (exidx_text_sno >= 0 && exidx_text_sno < num_sections) {
    const Shdr* exidx_text_shdr = &sections[exidx_text_sno];
    if (exidx_text_shdr->sh_size > 0) {
      uint32_t exidx_text_last_svma
        = exidx_text_shdr->sh_addr + exidx_text_shdr->sh_size - 1;
      exidx_text_last_avma
        = exidx_text_last_svma + text_bias;
    }
  }

  lul::ARMExToModule handler(smap, log);
  lul::ExceptionTableInfo
    parser(reinterpret_cast<const char*>(exidx_avma), exidx_size,
           reinterpret_cast<const char*>(extab_avma), extab_size,
           exidx_text_last_avma, &handler, log);
  parser.Start();
  return true;
}
#endif /* defined(LUL_PLAT_arm_android) */

bool LoadELF(const string& obj_file, MmapWrapper* map_wrapper,
             void** elf_header) {
  int obj_fd = open(obj_file.c_str(), O_RDONLY);
  if (obj_fd < 0) {
    fprintf(stderr, "Failed to open ELF file '%s': %s\n",
            obj_file.c_str(), strerror(errno));
    return false;
  }
  FDWrapper obj_fd_wrapper(obj_fd);
  struct stat st;
  if (fstat(obj_fd, &st) != 0 && st.st_size <= 0) {
    fprintf(stderr, "Unable to fstat ELF file '%s': %s\n",
            obj_file.c_str(), strerror(errno));
    return false;
  }
  // Mapping it read-only is good enough.  In any case, mapping it
  // read-write confuses Valgrind's debuginfo acquire/discard
  // heuristics, making it hard to profile the profiler.
  void *obj_base = mmap(nullptr, st.st_size,
                        PROT_READ, MAP_PRIVATE, obj_fd, 0);
  if (obj_base == MAP_FAILED) {
    fprintf(stderr, "Failed to mmap ELF file '%s': %s\n",
            obj_file.c_str(), strerror(errno));
    return false;
  }
  map_wrapper->set(obj_base, st.st_size);
  *elf_header = obj_base;
  if (!IsValidElf(*elf_header)) {
    fprintf(stderr, "Not a valid ELF file: %s\n", obj_file.c_str());
    return false;
  }
  return true;
}

// Get the endianness of ELF_HEADER. If it's invalid, return false.
template<typename ElfClass>
bool ElfEndianness(const typename ElfClass::Ehdr* elf_header,
                   bool* big_endian) {
  if (elf_header->e_ident[EI_DATA] == ELFDATA2LSB) {
    *big_endian = false;
    return true;
  }
  if (elf_header->e_ident[EI_DATA] == ELFDATA2MSB) {
    *big_endian = true;
    return true;
  }

  fprintf(stderr, "bad data encoding in ELF header: %d\n",
          elf_header->e_ident[EI_DATA]);
  return false;
}

//
// LoadSymbolsInfo
//
// Holds the state between the two calls to LoadSymbols() in case it's necessary
// to follow the .gnu_debuglink section and load debug information from a
// different file.
//
template<typename ElfClass>
class LoadSymbolsInfo {
 public:
  typedef typename ElfClass::Addr Addr;

  explicit LoadSymbolsInfo(const vector<string>& dbg_dirs) :
    debug_dirs_(dbg_dirs),
    has_loading_addr_(false) {}

  // Keeps track of which sections have been loaded so sections don't
  // accidentally get loaded twice from two different files.
  void LoadedSection(const string &section) {
    if (loaded_sections_.count(section) == 0) {
      loaded_sections_.insert(section);
    } else {
      fprintf(stderr, "Section %s has already been loaded.\n",
              section.c_str());
    }
  }

  string debuglink_file() const {
    return debuglink_file_;
  }

 private:
  const vector<string>& debug_dirs_; // Directories in which to
                                     // search for the debug ELF file.

  string debuglink_file_; // Full path to the debug ELF file.

  bool has_loading_addr_; // Indicate if LOADING_ADDR_ is valid.

  set<string> loaded_sections_; // Tracks the Loaded ELF sections
                                // between calls to LoadSymbols().
};

// Find the preferred loading address of the binary.
template<typename ElfClass>
typename ElfClass::Addr GetLoadingAddress(
    const typename ElfClass::Phdr* program_headers,
    int nheader) {
  typedef typename ElfClass::Phdr Phdr;

  // For non-PIC executables (e_type == ET_EXEC), the load address is
  // the start address of the first PT_LOAD segment.  (ELF requires
  // the segments to be sorted by load address.)  For PIC executables
  // and dynamic libraries (e_type == ET_DYN), this address will
  // normally be zero.
  for (int i = 0; i < nheader; ++i) {
    const Phdr& header = program_headers[i];
    if (header.p_type == PT_LOAD)
      return header.p_vaddr;
  }
  return 0;
}

template<typename ElfClass>
bool LoadSymbols(const string& obj_file,
                 const bool big_endian,
                 const typename ElfClass::Ehdr* elf_header,
                 const bool read_gnu_debug_link,
                 LoadSymbolsInfo<ElfClass>* info,
                 SecMap* smap,
                 void* rx_avma, size_t rx_size,
                 void (*log)(const char*)) {
  typedef typename ElfClass::Phdr Phdr;
  typedef typename ElfClass::Shdr Shdr;

  char buf[500];
  snprintf(buf, sizeof(buf), "LoadSymbols: BEGIN   %s\n", obj_file.c_str());
  buf[sizeof(buf)-1] = 0;
  log(buf);

  // This is how the text bias is calculated.
  // BEGIN CALCULATE BIAS
  uintptr_t loading_addr = GetLoadingAddress<ElfClass>(
      GetOffset<ElfClass, Phdr>(elf_header, elf_header->e_phoff),
      elf_header->e_phnum);
  uintptr_t text_bias = ((uintptr_t)rx_avma) - loading_addr;
  snprintf(buf, sizeof(buf),
           "LoadSymbols:   rx_avma=%llx, text_bias=%llx",
           (unsigned long long int)(uintptr_t)rx_avma,
           (unsigned long long int)text_bias);
  buf[sizeof(buf)-1] = 0;
  log(buf);
  // END CALCULATE BIAS

  const Shdr* sections =
      GetOffset<ElfClass, Shdr>(elf_header, elf_header->e_shoff);
  const Shdr* section_names = sections + elf_header->e_shstrndx;
  const char* names =
      GetOffset<ElfClass, char>(elf_header, section_names->sh_offset);
  const char *names_end = names + section_names->sh_size;
  bool found_usable_info = false;

  // Dwarf Call Frame Information (CFI) is actually independent from
  // the other DWARF debugging information, and can be used alone.
  const Shdr* dwarf_cfi_section =
      FindElfSectionByName<ElfClass>(".debug_frame", SHT_PROGBITS,
                                     sections, names, names_end,
                                     elf_header->e_shnum);
  if (dwarf_cfi_section) {
    // Ignore the return value of this function; even without call frame
    // information, the other debugging information could be perfectly
    // useful.
    info->LoadedSection(".debug_frame");
    bool result =
        LoadDwarfCFI<ElfClass>(obj_file, elf_header, ".debug_frame",
                               dwarf_cfi_section, false, 0, 0, big_endian,
                               smap, text_bias, log);
    found_usable_info = found_usable_info || result;
    if (result)
      log("LoadSymbols:   read CFI from .debug_frame");
  }

  // Linux C++ exception handling information can also provide
  // unwinding data.
  const Shdr* eh_frame_section =
      FindElfSectionByName<ElfClass>(".eh_frame", SHT_PROGBITS,
                                     sections, names, names_end,
                                     elf_header->e_shnum);
  if (eh_frame_section) {
    // Pointers in .eh_frame data may be relative to the base addresses of
    // certain sections. Provide those sections if present.
    const Shdr* got_section =
        FindElfSectionByName<ElfClass>(".got", SHT_PROGBITS,
                                       sections, names, names_end,
                                       elf_header->e_shnum);
    const Shdr* text_section =
        FindElfSectionByName<ElfClass>(".text", SHT_PROGBITS,
                                       sections, names, names_end,
                                       elf_header->e_shnum);
    info->LoadedSection(".eh_frame");
    // As above, ignore the return value of this function.
    bool result =
        LoadDwarfCFI<ElfClass>(obj_file, elf_header, ".eh_frame",
                               eh_frame_section, true,
                               got_section, text_section, big_endian,
                               smap, text_bias, log);
    found_usable_info = found_usable_info || result;
    if (result)
      log("LoadSymbols:   read CFI from .eh_frame");
  }

# if defined(LUL_PLAT_arm_android)
  // ARM has special unwind tables that can be used.  .exidx is
  // always required, and .extab is normally required, but may
  // be omitted if it is empty.  See comments on LoadARMexidx()
  // for more details.
  const Shdr* arm_exidx_section =
      FindElfSectionByName<ElfClass>(".ARM.exidx", SHT_ARM_EXIDX,
                                     sections, names, names_end,
                                     elf_header->e_shnum);
  const Shdr* arm_extab_section =
      FindElfSectionByName<ElfClass>(".ARM.extab", SHT_PROGBITS,
                                     sections, names, names_end,
                                     elf_header->e_shnum);
  const Shdr* debug_info_section =
      FindElfSectionByName<ElfClass>(".debug_info", SHT_PROGBITS,
                                     sections, names, names_end,
                                     elf_header->e_shnum);
  // Only load information from this section if there isn't a .debug_info
  // section.
  if (!debug_info_section && arm_exidx_section) {
    info->LoadedSection(".ARM.exidx");
    if (arm_extab_section)
      info->LoadedSection(".ARM.extab");
    bool result = LoadARMexidx<ElfClass>(elf_header,
                                         arm_exidx_section, arm_extab_section,
                                         text_bias,
                                         reinterpret_cast<uintptr_t>(rx_avma),
                                         rx_size, smap, log);
    found_usable_info = found_usable_info || result;
    if (result)
      log("LoadSymbols:   read EXIDX from .ARM.{exidx,extab}");
  }
# endif /* defined(LUL_PLAT_arm_android) */

  snprintf(buf, sizeof(buf), "LoadSymbols: END     %s\n", obj_file.c_str());
  buf[sizeof(buf)-1] = 0;
  log(buf);

  return found_usable_info;
}

// Return the breakpad symbol file identifier for the architecture of
// ELF_HEADER.
template<typename ElfClass>
const char* ElfArchitecture(const typename ElfClass::Ehdr* elf_header) {
  typedef typename ElfClass::Half Half;
  Half arch = elf_header->e_machine;
  switch (arch) {
    case EM_386:        return "x86";
    case EM_ARM:        return "arm";
    case EM_MIPS:       return "mips";
    case EM_PPC64:      return "ppc64";
    case EM_PPC:        return "ppc";
    case EM_S390:       return "s390";
    case EM_SPARC:      return "sparc";
    case EM_SPARCV9:    return "sparcv9";
    case EM_X86_64:     return "x86_64";
    default: return NULL;
  }
}

// Format the Elf file identifier in IDENTIFIER as a UUID with the
// dashes removed.
string FormatIdentifier(unsigned char identifier[16]) {
  char identifier_str[40];
  lul::FileID::ConvertIdentifierToString(
      identifier,
      identifier_str,
      sizeof(identifier_str));
  string id_no_dash;
  for (int i = 0; identifier_str[i] != '\0'; ++i)
    if (identifier_str[i] != '-')
      id_no_dash += identifier_str[i];
  // Add an extra "0" by the end.  PDB files on Windows have an 'age'
  // number appended to the end of the file identifier; this isn't
  // really used or necessary on other platforms, but be consistent.
  id_no_dash += '0';
  return id_no_dash;
}

// Return the non-directory portion of FILENAME: the portion after the
// last slash, or the whole filename if there are no slashes.
string BaseFileName(const string &filename) {
  // Lots of copies!  basename's behavior is less than ideal.
  char *c_filename = strdup(filename.c_str());
  string base = basename(c_filename);
  free(c_filename);
  return base;
}

template<typename ElfClass>
bool ReadSymbolDataElfClass(const typename ElfClass::Ehdr* elf_header,
                            const string& obj_filename,
                            const vector<string>& debug_dirs,
                            SecMap* smap, void* rx_avma, size_t rx_size,
                            void (*log)(const char*)) {
  typedef typename ElfClass::Ehdr Ehdr;

  unsigned char identifier[16];
  if (!lul
      ::FileID::ElfFileIdentifierFromMappedFile(elf_header, identifier)) {
    fprintf(stderr, "%s: unable to generate file identifier\n",
            obj_filename.c_str());
    return false;
  }

  const char *architecture = ElfArchitecture<ElfClass>(elf_header);
  if (!architecture) {
    fprintf(stderr, "%s: unrecognized ELF machine architecture: %d\n",
            obj_filename.c_str(), elf_header->e_machine);
    return false;
  }

  // Figure out what endianness this file is.
  bool big_endian;
  if (!ElfEndianness<ElfClass>(elf_header, &big_endian))
    return false;

  string name = BaseFileName(obj_filename);
  string os = "Linux";
  string id = FormatIdentifier(identifier);

  LoadSymbolsInfo<ElfClass> info(debug_dirs);
  if (!LoadSymbols<ElfClass>(obj_filename, big_endian, elf_header,
                             !debug_dirs.empty(), &info,
                             smap, rx_avma, rx_size, log)) {
    const string debuglink_file = info.debuglink_file();
    if (debuglink_file.empty())
      return false;

    // Load debuglink ELF file.
    fprintf(stderr, "Found debugging info in %s\n", debuglink_file.c_str());
    MmapWrapper debug_map_wrapper;
    Ehdr* debug_elf_header = NULL;
    if (!LoadELF(debuglink_file, &debug_map_wrapper,
                 reinterpret_cast<void**>(&debug_elf_header)))
      return false;
    // Sanity checks to make sure everything matches up.
    const char *debug_architecture =
        ElfArchitecture<ElfClass>(debug_elf_header);
    if (!debug_architecture) {
      fprintf(stderr, "%s: unrecognized ELF machine architecture: %d\n",
              debuglink_file.c_str(), debug_elf_header->e_machine);
      return false;
    }
    if (strcmp(architecture, debug_architecture)) {
      fprintf(stderr, "%s with ELF machine architecture %s does not match "
              "%s with ELF architecture %s\n",
              debuglink_file.c_str(), debug_architecture,
              obj_filename.c_str(), architecture);
      return false;
    }

    bool debug_big_endian;
    if (!ElfEndianness<ElfClass>(debug_elf_header, &debug_big_endian))
      return false;
    if (debug_big_endian != big_endian) {
      fprintf(stderr, "%s and %s does not match in endianness\n",
              obj_filename.c_str(), debuglink_file.c_str());
      return false;
    }

    if (!LoadSymbols<ElfClass>(debuglink_file, debug_big_endian,
                               debug_elf_header, false, &info,
                               smap, rx_avma, rx_size, log)) {
      return false;
    }
  }

  return true;
}

}  // namespace (anon)


namespace lul {

bool ReadSymbolDataInternal(const uint8_t* obj_file,
                            const string& obj_filename,
                            const vector<string>& debug_dirs,
                            SecMap* smap, void* rx_avma, size_t rx_size,
                            void (*log)(const char*)) {

  if (!IsValidElf(obj_file)) {
    fprintf(stderr, "Not a valid ELF file: %s\n", obj_filename.c_str());
    return false;
  }

  int elfclass = ElfClass(obj_file);
  if (elfclass == ELFCLASS32) {
    return ReadSymbolDataElfClass<ElfClass32>(
        reinterpret_cast<const Elf32_Ehdr*>(obj_file),
        obj_filename, debug_dirs, smap, rx_avma, rx_size, log);
  }
  if (elfclass == ELFCLASS64) {
    return ReadSymbolDataElfClass<ElfClass64>(
        reinterpret_cast<const Elf64_Ehdr*>(obj_file),
        obj_filename, debug_dirs, smap, rx_avma, rx_size, log);
  }

  return false;
}

bool ReadSymbolData(const string& obj_file,
                    const vector<string>& debug_dirs,
                    SecMap* smap, void* rx_avma, size_t rx_size,
                    void (*log)(const char*)) {
  MmapWrapper map_wrapper;
  void* elf_header = NULL;
  if (!LoadELF(obj_file, &map_wrapper, &elf_header))
    return false;

  return ReadSymbolDataInternal(reinterpret_cast<uint8_t*>(elf_header),
                                obj_file, debug_dirs,
                                smap, rx_avma, rx_size, log);
}


namespace {

template<typename ElfClass>
void FindElfClassSection(const char *elf_base,
                         const char *section_name,
                         typename ElfClass::Word section_type,
                         const void **section_start,
                         int *section_size) {
  typedef typename ElfClass::Ehdr Ehdr;
  typedef typename ElfClass::Shdr Shdr;

  MOZ_ASSERT(elf_base);
  MOZ_ASSERT(section_start);
  MOZ_ASSERT(section_size);

  MOZ_ASSERT(strncmp(elf_base, ELFMAG, SELFMAG) == 0);

  const Ehdr* elf_header = reinterpret_cast<const Ehdr*>(elf_base);
  MOZ_ASSERT(elf_header->e_ident[EI_CLASS] == ElfClass::kClass);

  const Shdr* sections =
    GetOffset<ElfClass,Shdr>(elf_header, elf_header->e_shoff);
  const Shdr* section_names = sections + elf_header->e_shstrndx;
  const char* names =
    GetOffset<ElfClass,char>(elf_header, section_names->sh_offset);
  const char *names_end = names + section_names->sh_size;

  const Shdr* section =
    FindElfSectionByName<ElfClass>(section_name, section_type,
                                   sections, names, names_end,
                                   elf_header->e_shnum);

  if (section != NULL && section->sh_size > 0) {
    *section_start = elf_base + section->sh_offset;
    *section_size = section->sh_size;
  }
}

template<typename ElfClass>
void FindElfClassSegment(const char *elf_base,
                         typename ElfClass::Word segment_type,
                         const void **segment_start,
                         int *segment_size) {
  typedef typename ElfClass::Ehdr Ehdr;
  typedef typename ElfClass::Phdr Phdr;

  MOZ_ASSERT(elf_base);
  MOZ_ASSERT(segment_start);
  MOZ_ASSERT(segment_size);

  MOZ_ASSERT(strncmp(elf_base, ELFMAG, SELFMAG) == 0);

  const Ehdr* elf_header = reinterpret_cast<const Ehdr*>(elf_base);
  MOZ_ASSERT(elf_header->e_ident[EI_CLASS] == ElfClass::kClass);

  const Phdr* phdrs =
    GetOffset<ElfClass,Phdr>(elf_header, elf_header->e_phoff);

  for (int i = 0; i < elf_header->e_phnum; ++i) {
    if (phdrs[i].p_type == segment_type) {
      *segment_start = elf_base + phdrs[i].p_offset;
      *segment_size = phdrs[i].p_filesz;
      return;
    }
  }
}

}  // namespace (anon)

bool IsValidElf(const void* elf_base) {
  return strncmp(reinterpret_cast<const char*>(elf_base),
                 ELFMAG, SELFMAG) == 0;
}

int ElfClass(const void* elf_base) {
  const ElfW(Ehdr)* elf_header =
    reinterpret_cast<const ElfW(Ehdr)*>(elf_base);

  return elf_header->e_ident[EI_CLASS];
}

bool FindElfSection(const void *elf_mapped_base,
                    const char *section_name,
                    uint32_t section_type,
                    const void **section_start,
                    int *section_size,
                    int *elfclass) {
  MOZ_ASSERT(elf_mapped_base);
  MOZ_ASSERT(section_start);
  MOZ_ASSERT(section_size);

  *section_start = NULL;
  *section_size = 0;

  if (!IsValidElf(elf_mapped_base))
    return false;

  int cls = ElfClass(elf_mapped_base);
  if (elfclass) {
    *elfclass = cls;
  }

  const char* elf_base =
    static_cast<const char*>(elf_mapped_base);

  if (cls == ELFCLASS32) {
    FindElfClassSection<ElfClass32>(elf_base, section_name, section_type,
                                    section_start, section_size);
    return *section_start != NULL;
  } else if (cls == ELFCLASS64) {
    FindElfClassSection<ElfClass64>(elf_base, section_name, section_type,
                                    section_start, section_size);
    return *section_start != NULL;
  }

  return false;
}

bool FindElfSegment(const void *elf_mapped_base,
                    uint32_t segment_type,
                    const void **segment_start,
                    int *segment_size,
                    int *elfclass) {
  MOZ_ASSERT(elf_mapped_base);
  MOZ_ASSERT(segment_start);
  MOZ_ASSERT(segment_size);

  *segment_start = NULL;
  *segment_size = 0;

  if (!IsValidElf(elf_mapped_base))
    return false;

  int cls = ElfClass(elf_mapped_base);
  if (elfclass) {
    *elfclass = cls;
  }

  const char* elf_base =
    static_cast<const char*>(elf_mapped_base);

  if (cls == ELFCLASS32) {
    FindElfClassSegment<ElfClass32>(elf_base, segment_type,
                                    segment_start, segment_size);
    return *segment_start != NULL;
  } else if (cls == ELFCLASS64) {
    FindElfClassSegment<ElfClass64>(elf_base, segment_type,
                                    segment_start, segment_size);
    return *segment_start != NULL;
  }

  return false;
}


// (derived from)
// file_id.cc: Return a unique identifier for a file
//
// See file_id.h for documentation
//

// ELF note name and desc are 32-bits word padded.
#define NOTE_PADDING(a) ((a + 3) & ~3)

// These functions are also used inside the crashed process, so be safe
// and use the syscall/libc wrappers instead of direct syscalls or libc.

template<typename ElfClass>
static bool ElfClassBuildIDNoteIdentifier(const void *section, int length,
                                          uint8_t identifier[kMDGUIDSize]) {
  typedef typename ElfClass::Nhdr Nhdr;

  const void* section_end = reinterpret_cast<const char*>(section) + length;
  const Nhdr* note_header = reinterpret_cast<const Nhdr*>(section);
  while (reinterpret_cast<const void *>(note_header) < section_end) {
    if (note_header->n_type == NT_GNU_BUILD_ID)
      break;
    note_header = reinterpret_cast<const Nhdr*>(
                  reinterpret_cast<const char*>(note_header) + sizeof(Nhdr) +
                  NOTE_PADDING(note_header->n_namesz) +
                  NOTE_PADDING(note_header->n_descsz));
  }
  if (reinterpret_cast<const void *>(note_header) >= section_end ||
      note_header->n_descsz == 0) {
    return false;
  }

  const char* build_id = reinterpret_cast<const char*>(note_header) +
    sizeof(Nhdr) + NOTE_PADDING(note_header->n_namesz);
  // Copy as many bits of the build ID as will fit
  // into the GUID space.
  memset(identifier, 0, kMDGUIDSize);
  memcpy(identifier, build_id,
         std::min(kMDGUIDSize, (size_t)note_header->n_descsz));

  return true;
}

// Attempt to locate a .note.gnu.build-id section in an ELF binary
// and copy as many bytes of it as will fit into |identifier|.
static bool FindElfBuildIDNote(const void *elf_mapped_base,
                               uint8_t identifier[kMDGUIDSize]) {
  void* note_section;
  int note_size, elfclass;
  if ((!FindElfSegment(elf_mapped_base, PT_NOTE,
                       (const void**)&note_section, &note_size, &elfclass) ||
      note_size == 0)  &&
      (!FindElfSection(elf_mapped_base, ".note.gnu.build-id", SHT_NOTE,
                       (const void**)&note_section, &note_size, &elfclass) ||
      note_size == 0)) {
    return false;
  }

  if (elfclass == ELFCLASS32) {
    return ElfClassBuildIDNoteIdentifier<ElfClass32>(note_section, note_size,
                                                     identifier);
  } else if (elfclass == ELFCLASS64) {
    return ElfClassBuildIDNoteIdentifier<ElfClass64>(note_section, note_size,
                                                     identifier);
  }

  return false;
}

// Attempt to locate the .text section of an ELF binary and generate
// a simple hash by XORing the first page worth of bytes into |identifier|.
static bool HashElfTextSection(const void *elf_mapped_base,
                               uint8_t identifier[kMDGUIDSize]) {
  void* text_section;
  int text_size;
  if (!FindElfSection(elf_mapped_base, ".text", SHT_PROGBITS,
                      (const void**)&text_section, &text_size, NULL) ||
      text_size == 0) {
    return false;
  }

  memset(identifier, 0, kMDGUIDSize);
  const uint8_t* ptr = reinterpret_cast<const uint8_t*>(text_section);
  const uint8_t* ptr_end = ptr + std::min(text_size, 4096);
  while (ptr < ptr_end) {
    for (unsigned i = 0; i < kMDGUIDSize; i++)
      identifier[i] ^= ptr[i];
    ptr += kMDGUIDSize;
  }
  return true;
}

// static
bool FileID::ElfFileIdentifierFromMappedFile(const void* base,
                                             uint8_t identifier[kMDGUIDSize]) {
  // Look for a build id note first.
  if (FindElfBuildIDNote(base, identifier))
    return true;

  // Fall back on hashing the first page of the text section.
  return HashElfTextSection(base, identifier);
}

// static
void FileID::ConvertIdentifierToString(const uint8_t identifier[kMDGUIDSize],
                                       char* buffer, int buffer_length) {
  uint8_t identifier_swapped[kMDGUIDSize];

  // Endian-ness swap to match dump processor expectation.
  memcpy(identifier_swapped, identifier, kMDGUIDSize);
  uint32_t* data1 = reinterpret_cast<uint32_t*>(identifier_swapped);
  *data1 = htonl(*data1);
  uint16_t* data2 = reinterpret_cast<uint16_t*>(identifier_swapped + 4);
  *data2 = htons(*data2);
  uint16_t* data3 = reinterpret_cast<uint16_t*>(identifier_swapped + 6);
  *data3 = htons(*data3);

  int buffer_idx = 0;
  for (unsigned int idx = 0;
       (buffer_idx < buffer_length) && (idx < kMDGUIDSize);
       ++idx) {
    int hi = (identifier_swapped[idx] >> 4) & 0x0F;
    int lo = (identifier_swapped[idx]) & 0x0F;

    if (idx == 4 || idx == 6 || idx == 8 || idx == 10)
      buffer[buffer_idx++] = '-';

    buffer[buffer_idx++] = (hi >= 10) ? 'A' + hi - 10 : '0' + hi;
    buffer[buffer_idx++] = (lo >= 10) ? 'A' + lo - 10 : '0' + lo;
  }

  // NULL terminate
  buffer[(buffer_idx < buffer_length) ? buffer_idx : buffer_idx - 1] = 0;
}

}  // namespace lul