static ABSL_ATTRIBUTE_NOINLINE FindSymbolResult FindSymbol()

in absl/debugging/symbolize_elf.inc [662:774]
71 lines of code
23 McCabe index (conditional complexity)

static ABSL_ATTRIBUTE_NOINLINE FindSymbolResult FindSymbol(
    const void *const pc, const int fd, char *out, int out_size,
    ptrdiff_t relocation, const ElfW(Shdr) * strtab, const ElfW(Shdr) * symtab,
    const ElfW(Shdr) * opd, char *tmp_buf, int tmp_buf_size) {
  if (symtab == nullptr) {
    return SYMBOL_NOT_FOUND;
  }

  // Read multiple symbols at once to save read() calls.
  ElfW(Sym) *buf = reinterpret_cast<ElfW(Sym) *>(tmp_buf);
  const int buf_entries = tmp_buf_size / sizeof(buf[0]);

  const int num_symbols = symtab->sh_size / symtab->sh_entsize;

  // On platforms using an .opd section (PowerPC & IA64), a function symbol
  // has the address of a function descriptor, which contains the real
  // starting address.  However, we do not always want to use the real
  // starting address because we sometimes want to symbolize a function
  // pointer into the .opd section, e.g. FindSymbol(&foo,...).
  const bool pc_in_opd =
      kPlatformUsesOPDSections && opd != nullptr && InSection(pc, opd);
  const bool deref_function_descriptor_pointer =
      kPlatformUsesOPDSections && opd != nullptr && !pc_in_opd;

  ElfW(Sym) best_match;
  SafeMemZero(&best_match, sizeof(best_match));
  bool found_match = false;
  for (int i = 0; i < num_symbols;) {
    off_t offset = symtab->sh_offset + i * symtab->sh_entsize;
    const int num_remaining_symbols = num_symbols - i;
    const int entries_in_chunk = std::min(num_remaining_symbols, buf_entries);
    const int bytes_in_chunk = entries_in_chunk * sizeof(buf[0]);
    const ssize_t len = ReadFromOffset(fd, buf, bytes_in_chunk, offset);
    SAFE_ASSERT(len % sizeof(buf[0]) == 0);
    const ssize_t num_symbols_in_buf = len / sizeof(buf[0]);
    SAFE_ASSERT(num_symbols_in_buf <= entries_in_chunk);
    for (int j = 0; j < num_symbols_in_buf; ++j) {
      const ElfW(Sym) &symbol = buf[j];

      // For a DSO, a symbol address is relocated by the loading address.
      // We keep the original address for opd redirection below.
      const char *const original_start_address =
          reinterpret_cast<const char *>(symbol.st_value);
      const char *start_address =
          ComputeOffset(original_start_address, relocation);

#ifdef __arm__
      // ARM functions are always aligned to multiples of two bytes; the
      // lowest-order bit in start_address is ignored by the CPU and indicates
      // whether the function contains ARM (0) or Thumb (1) code. We don't care
      // about what encoding is being used; we just want the real start address
      // of the function.
      start_address = reinterpret_cast<const char *>(
          reinterpret_cast<uintptr_t>(start_address) & ~1);
#endif

      if (deref_function_descriptor_pointer &&
          InSection(original_start_address, opd)) {
        // The opd section is mapped into memory.  Just dereference
        // start_address to get the first double word, which points to the
        // function entry.
        start_address = *reinterpret_cast<const char *const *>(start_address);
      }

      // If pc is inside the .opd section, it points to a function descriptor.
      const size_t size = pc_in_opd ? kFunctionDescriptorSize : symbol.st_size;
      const void *const end_address = ComputeOffset(start_address, size);
      if (symbol.st_value != 0 &&  // Skip null value symbols.
          symbol.st_shndx != 0 &&  // Skip undefined symbols.
#ifdef STT_TLS
          ELF_ST_TYPE(symbol.st_info) != STT_TLS &&  // Skip thread-local data.
#endif                                               // STT_TLS
          ((start_address <= pc && pc < end_address) ||
           (start_address == pc && pc == end_address))) {
        if (!found_match || ShouldPickFirstSymbol(symbol, best_match)) {
          found_match = true;
          best_match = symbol;
        }
      }
    }
    i += num_symbols_in_buf;
  }

  if (found_match) {
    const size_t off = strtab->sh_offset + best_match.st_name;
    const ssize_t n_read = ReadFromOffset(fd, out, out_size, off);
    if (n_read <= 0) {
      // This should never happen.
      ABSL_RAW_LOG(WARNING,
                   "Unable to read from fd %d at offset %zu: n_read = %zd", fd,
                   off, n_read);
      return SYMBOL_NOT_FOUND;
    }
    ABSL_RAW_CHECK(n_read <= out_size, "ReadFromOffset read too much data.");

    // strtab->sh_offset points into .strtab-like section that contains
    // NUL-terminated strings: '\0foo\0barbaz\0...".
    //
    // sh_offset+st_name points to the start of symbol name, but we don't know
    // how long the symbol is, so we try to read as much as we have space for,
    // and usually over-read (i.e. there is a NUL somewhere before n_read).
    if (memchr(out, '\0', n_read) == nullptr) {
      // Either out_size was too small (n_read == out_size and no NUL), or
      // we tried to read past the EOF (n_read < out_size) and .strtab is
      // corrupt (missing terminating NUL; should never happen for valid ELF).
      out[n_read - 1] = '\0';
      return SYMBOL_TRUNCATED;
    }
    return SYMBOL_FOUND;
  }

  return SYMBOL_NOT_FOUND;
}