Image.cpp 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455
  1. /*
  2. * Copyright (c) 2018-2021, Andreas Kling <kling@serenityos.org>
  3. * Copyright (c) 2022, the SerenityOS developers.
  4. *
  5. * SPDX-License-Identifier: BSD-2-Clause
  6. */
  7. #include <AK/BinarySearch.h>
  8. #include <AK/Debug.h>
  9. #include <AK/Demangle.h>
  10. #include <AK/Memory.h>
  11. #include <AK/QuickSort.h>
  12. #include <AK/StringBuilder.h>
  13. #include <AK/StringView.h>
  14. #include <LibELF/Image.h>
  15. #include <LibELF/Validation.h>
  16. #include <limits.h>
  17. namespace ELF {
  18. Image::Image(ReadonlyBytes bytes, bool verbose_logging)
  19. : m_buffer(bytes.data())
  20. , m_size(bytes.size())
  21. , m_verbose_logging(verbose_logging)
  22. {
  23. parse();
  24. }
  25. Image::Image(u8 const* buffer, size_t size, bool verbose_logging)
  26. : Image(ReadonlyBytes { buffer, size }, verbose_logging)
  27. {
  28. }
  29. StringView Image::section_index_to_string(unsigned index) const
  30. {
  31. VERIFY(m_valid);
  32. if (index == SHN_UNDEF)
  33. return "Undefined"sv;
  34. if (index >= SHN_LORESERVE)
  35. return "Reserved"sv;
  36. return section(index).name();
  37. }
  38. unsigned Image::symbol_count() const
  39. {
  40. VERIFY(m_valid);
  41. if (!section_count())
  42. return 0;
  43. return section(m_symbol_table_section_index).entry_count();
  44. }
  45. void Image::dump() const
  46. {
  47. #if ELF_IMAGE_DEBUG
  48. dbgln("ELF::Image({:p}) {{", this);
  49. dbgln(" is_valid: {}", is_valid());
  50. if (!is_valid()) {
  51. dbgln("}}");
  52. return;
  53. }
  54. dbgln(" type: {}", ELF::Image::object_file_type_to_string(header().e_type).value_or("(?)"sv));
  55. dbgln(" machine: {}", header().e_machine);
  56. dbgln(" entry: {:x}", header().e_entry);
  57. dbgln(" shoff: {}", header().e_shoff);
  58. dbgln(" shnum: {}", header().e_shnum);
  59. dbgln(" phoff: {}", header().e_phoff);
  60. dbgln(" phnum: {}", header().e_phnum);
  61. dbgln(" shstrndx: {}", header().e_shstrndx);
  62. for_each_program_header([&](ProgramHeader const& program_header) {
  63. dbgln(" Program Header {}: {{", program_header.index());
  64. dbgln(" type: {:x}", program_header.type());
  65. dbgln(" offset: {:x}", program_header.offset());
  66. dbgln(" flags: {:x}", program_header.flags());
  67. dbgln(" }}");
  68. });
  69. for (unsigned i = 0; i < header().e_shnum; ++i) {
  70. auto const& section = this->section(i);
  71. dbgln(" Section {}: {{", i);
  72. dbgln(" name: {}", section.name());
  73. dbgln(" type: {:x}", section.type());
  74. dbgln(" offset: {:x}", section.offset());
  75. dbgln(" size: {}", section.size());
  76. dbgln(" ");
  77. dbgln(" }}");
  78. }
  79. dbgln("Symbol count: {} (table is {})", symbol_count(), m_symbol_table_section_index);
  80. for (unsigned i = 1; i < symbol_count(); ++i) {
  81. auto const& sym = symbol(i);
  82. dbgln("Symbol @{}:", i);
  83. dbgln(" Name: {}", sym.name());
  84. dbgln(" In section: {}", section_index_to_string(sym.section_index()));
  85. dbgln(" Value: {}", sym.value());
  86. dbgln(" Size: {}", sym.size());
  87. }
  88. dbgln("}}");
  89. #endif
  90. }
  91. unsigned Image::section_count() const
  92. {
  93. VERIFY(m_valid);
  94. return header().e_shnum;
  95. }
  96. unsigned Image::program_header_count() const
  97. {
  98. VERIFY(m_valid);
  99. return header().e_phnum;
  100. }
  101. bool Image::parse()
  102. {
  103. if (m_size < sizeof(ElfW(Ehdr)) || !validate_elf_header(header(), m_size, m_verbose_logging)) {
  104. if (m_verbose_logging)
  105. dbgln("ELF::Image::parse(): ELF Header not valid");
  106. m_valid = false;
  107. return false;
  108. }
  109. auto result_or_error = validate_program_headers(header(), m_size, { m_buffer, m_size }, nullptr, m_verbose_logging);
  110. if (result_or_error.is_error()) {
  111. if (m_verbose_logging)
  112. dbgln("ELF::Image::parse(): Failed validating ELF Program Headers");
  113. m_valid = false;
  114. return false;
  115. }
  116. if (!result_or_error.value()) {
  117. if (m_verbose_logging)
  118. dbgln("ELF::Image::parse(): ELF Program Headers not valid");
  119. m_valid = false;
  120. return false;
  121. }
  122. m_valid = true;
  123. // First locate the string tables.
  124. for (unsigned i = 0; i < section_count(); ++i) {
  125. auto& sh = section_header(i);
  126. if (sh.sh_type == SHT_SYMTAB) {
  127. if (m_symbol_table_section_index && m_symbol_table_section_index != i) {
  128. m_valid = false;
  129. return false;
  130. }
  131. m_symbol_table_section_index = i;
  132. }
  133. if (sh.sh_type == SHT_STRTAB && i != header().e_shstrndx) {
  134. if (section_header_table_string(sh.sh_name) == ELF_STRTAB)
  135. m_string_table_section_index = i;
  136. }
  137. }
  138. return m_valid;
  139. }
  140. StringView Image::table_string(unsigned table_index, unsigned offset) const
  141. {
  142. VERIFY(m_valid);
  143. auto& sh = section_header(table_index);
  144. if (sh.sh_type != SHT_STRTAB)
  145. return {};
  146. size_t computed_offset = sh.sh_offset + offset;
  147. if (computed_offset >= m_size) {
  148. if (m_verbose_logging)
  149. dbgln("SHENANIGANS! Image::table_string() computed offset outside image.");
  150. return {};
  151. }
  152. size_t max_length = min(m_size - computed_offset, (size_t)PAGE_SIZE);
  153. size_t length = strnlen(raw_data(sh.sh_offset + offset), max_length);
  154. return { raw_data(sh.sh_offset + offset), length };
  155. }
  156. StringView Image::section_header_table_string(unsigned offset) const
  157. {
  158. VERIFY(m_valid);
  159. return table_string(header().e_shstrndx, offset);
  160. }
  161. StringView Image::table_string(unsigned offset) const
  162. {
  163. VERIFY(m_valid);
  164. return table_string(m_string_table_section_index, offset);
  165. }
  166. char const* Image::raw_data(unsigned offset) const
  167. {
  168. VERIFY(offset < m_size); // Callers must check indices into raw_data()'s result are also in bounds.
  169. return reinterpret_cast<char const*>(m_buffer) + offset;
  170. }
  171. const ElfW(Ehdr) & Image::header() const
  172. {
  173. VERIFY(m_size >= sizeof(ElfW(Ehdr)));
  174. return *reinterpret_cast<const ElfW(Ehdr)*>(raw_data(0));
  175. }
  176. const ElfW(Phdr) & Image::program_header_internal(unsigned index) const
  177. {
  178. VERIFY(m_valid);
  179. VERIFY(index < header().e_phnum);
  180. return *reinterpret_cast<const ElfW(Phdr)*>(raw_data(header().e_phoff + (index * sizeof(ElfW(Phdr)))));
  181. }
  182. const ElfW(Shdr) & Image::section_header(unsigned index) const
  183. {
  184. VERIFY(m_valid);
  185. VERIFY(index < header().e_shnum);
  186. return *reinterpret_cast<const ElfW(Shdr)*>(raw_data(header().e_shoff + (index * header().e_shentsize)));
  187. }
  188. Image::Symbol Image::symbol(unsigned index) const
  189. {
  190. VERIFY(m_valid);
  191. VERIFY(index < symbol_count());
  192. auto* raw_syms = reinterpret_cast<const ElfW(Sym)*>(raw_data(section(m_symbol_table_section_index).offset()));
  193. return Symbol(*this, index, raw_syms[index]);
  194. }
  195. Image::Section Image::section(unsigned index) const
  196. {
  197. VERIFY(m_valid);
  198. VERIFY(index < section_count());
  199. return Section(*this, index);
  200. }
  201. Image::ProgramHeader Image::program_header(unsigned index) const
  202. {
  203. VERIFY(m_valid);
  204. VERIFY(index < program_header_count());
  205. return ProgramHeader(*this, index);
  206. }
  207. Image::Relocation Image::RelocationSection::relocation(unsigned index) const
  208. {
  209. VERIFY(index < relocation_count());
  210. auto* rels = reinterpret_cast<const ElfW(Rel)*>(m_image.raw_data(offset()));
  211. return Relocation(m_image, rels[index]);
  212. }
  213. Optional<Image::RelocationSection> Image::Section::relocations() const
  214. {
  215. StringBuilder builder;
  216. builder.append(".rel"sv);
  217. builder.append(name());
  218. auto relocation_section = m_image.lookup_section(builder.string_view());
  219. if (!relocation_section.has_value())
  220. return {};
  221. dbgln_if(ELF_IMAGE_DEBUG, "Found relocations for {} in {}", name(), relocation_section.value().name());
  222. return static_cast<RelocationSection>(relocation_section.value());
  223. }
  224. Optional<Image::Section> Image::lookup_section(StringView name) const
  225. {
  226. VERIFY(m_valid);
  227. for (unsigned i = 0; i < section_count(); ++i) {
  228. auto section = this->section(i);
  229. if (section.name() == name)
  230. return section;
  231. }
  232. return {};
  233. }
  234. Optional<StringView> Image::object_file_type_to_string(ElfW(Half) type)
  235. {
  236. switch (type) {
  237. case ET_NONE:
  238. return "None"sv;
  239. case ET_REL:
  240. return "Relocatable"sv;
  241. case ET_EXEC:
  242. return "Executable"sv;
  243. case ET_DYN:
  244. return "Shared object"sv;
  245. case ET_CORE:
  246. return "Core"sv;
  247. default:
  248. return {};
  249. }
  250. }
  251. Optional<StringView> Image::object_machine_type_to_string(ElfW(Half) type)
  252. {
  253. switch (type) {
  254. case ET_NONE:
  255. return "None"sv;
  256. case EM_M32:
  257. return "AT&T WE 32100"sv;
  258. case EM_SPARC:
  259. return "SPARC"sv;
  260. case EM_386:
  261. return "Intel 80386"sv;
  262. case EM_68K:
  263. return "Motorola 68000"sv;
  264. case EM_88K:
  265. return "Motorola 88000"sv;
  266. case EM_486:
  267. return "Intel 80486"sv;
  268. case EM_860:
  269. return "Intel 80860"sv;
  270. case EM_MIPS:
  271. return "MIPS R3000 Big-Endian only"sv;
  272. case EM_X86_64:
  273. return "x86_64"sv;
  274. default:
  275. return {};
  276. }
  277. }
  278. Optional<StringView> Image::object_abi_type_to_string(Elf_Byte type)
  279. {
  280. switch (type) {
  281. case ELFOSABI_SYSV:
  282. return "SYSV"sv;
  283. case ELFOSABI_HPUX:
  284. return "HP-UX"sv;
  285. case ELFOSABI_NETBSD:
  286. return "NetBSD"sv;
  287. case ELFOSABI_LINUX:
  288. return "Linux"sv;
  289. case ELFOSABI_HURD:
  290. return "GNU Hurd"sv;
  291. case ELFOSABI_86OPEN:
  292. return "86Open"sv;
  293. case ELFOSABI_SOLARIS:
  294. return "Solaris"sv;
  295. case ELFOSABI_MONTEREY:
  296. return "AIX"sv;
  297. case ELFOSABI_IRIX:
  298. return "IRIX"sv;
  299. case ELFOSABI_FREEBSD:
  300. return "FreeBSD"sv;
  301. case ELFOSABI_TRU64:
  302. return "Tru64"sv;
  303. case ELFOSABI_MODESTO:
  304. return "Novell Modesto"sv;
  305. case ELFOSABI_OPENBSD:
  306. return "OpenBSD"sv;
  307. case ELFOSABI_ARM:
  308. return "ARM"sv;
  309. case ELFOSABI_STANDALONE:
  310. return "Standalone"sv;
  311. default:
  312. return {};
  313. }
  314. }
  315. StringView Image::Symbol::raw_data() const
  316. {
  317. auto section = this->section();
  318. return { section.raw_data() + (value() - section.address()), size() };
  319. }
  320. #ifndef KERNEL
  321. Optional<Image::Symbol> Image::find_demangled_function(StringView name) const
  322. {
  323. Optional<Image::Symbol> found;
  324. for_each_symbol([&](Image::Symbol const& symbol) {
  325. if (symbol.type() != STT_FUNC && symbol.type() != STT_GNU_IFUNC)
  326. return IterationDecision::Continue;
  327. if (symbol.is_undefined())
  328. return IterationDecision::Continue;
  329. auto demangled = demangle(symbol.name());
  330. auto index_of_paren = demangled.find('(');
  331. if (index_of_paren.has_value()) {
  332. demangled = demangled.substring(0, index_of_paren.value());
  333. }
  334. if (demangled != name)
  335. return IterationDecision::Continue;
  336. found = symbol;
  337. return IterationDecision::Break;
  338. });
  339. return found;
  340. }
  341. Image::SortedSymbol* Image::find_sorted_symbol(FlatPtr address) const
  342. {
  343. if (m_sorted_symbols.is_empty())
  344. sort_symbols();
  345. size_t index = 0;
  346. binary_search(m_sorted_symbols, nullptr, &index, [&address](auto, auto& candidate) {
  347. if (address < candidate.address)
  348. return -1;
  349. else if (address > candidate.address)
  350. return 1;
  351. else
  352. return 0;
  353. });
  354. // FIXME: The error path here feels strange, index == 0 means error but what about symbol #0?
  355. if (index == 0)
  356. return nullptr;
  357. return &m_sorted_symbols[index];
  358. }
  359. Optional<Image::Symbol> Image::find_symbol(FlatPtr address, u32* out_offset) const
  360. {
  361. auto symbol_count = this->symbol_count();
  362. if (!symbol_count)
  363. return {};
  364. auto* symbol = find_sorted_symbol(address);
  365. if (!symbol)
  366. return {};
  367. if (out_offset)
  368. *out_offset = address - symbol->address;
  369. return symbol->symbol;
  370. }
  371. NEVER_INLINE void Image::sort_symbols() const
  372. {
  373. m_sorted_symbols.ensure_capacity(symbol_count());
  374. for_each_symbol([this](auto const& symbol) {
  375. m_sorted_symbols.append({ symbol.value(), symbol.name(), {}, symbol });
  376. });
  377. quick_sort(m_sorted_symbols, [](auto& a, auto& b) {
  378. return a.address < b.address;
  379. });
  380. }
  381. String Image::symbolicate(FlatPtr address, u32* out_offset) const
  382. {
  383. auto symbol_count = this->symbol_count();
  384. if (!symbol_count) {
  385. if (out_offset)
  386. *out_offset = 0;
  387. return "??";
  388. }
  389. auto* symbol = find_sorted_symbol(address);
  390. if (!symbol) {
  391. if (out_offset)
  392. *out_offset = 0;
  393. return "??";
  394. }
  395. auto& demangled_name = symbol->demangled_name;
  396. if (demangled_name.is_null())
  397. demangled_name = demangle(symbol->name);
  398. if (out_offset) {
  399. *out_offset = address - symbol->address;
  400. return demangled_name;
  401. }
  402. return String::formatted("{} +{:#x}", demangled_name, address - symbol->address);
  403. }
  404. #endif
  405. } // end namespace ELF