Image.cpp 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453
  1. /*
  2. * Copyright (c) 2018-2021, Andreas Kling <kling@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <AK/BinarySearch.h>
  7. #include <AK/Debug.h>
  8. #include <AK/Demangle.h>
  9. #include <AK/Memory.h>
  10. #include <AK/QuickSort.h>
  11. #include <AK/StringBuilder.h>
  12. #include <AK/StringView.h>
  13. #include <LibELF/Image.h>
  14. #include <LibELF/Validation.h>
  15. #include <limits.h>
  16. namespace ELF {
  17. Image::Image(ReadonlyBytes bytes, bool verbose_logging)
  18. : m_buffer(bytes.data())
  19. , m_size(bytes.size())
  20. , m_verbose_logging(verbose_logging)
  21. {
  22. parse();
  23. }
  24. Image::Image(const u8* buffer, size_t size, bool verbose_logging)
  25. : Image(ReadonlyBytes { buffer, size }, verbose_logging)
  26. {
  27. }
  28. Image::~Image()
  29. {
  30. }
  31. StringView Image::section_index_to_string(unsigned index) const
  32. {
  33. VERIFY(m_valid);
  34. if (index == SHN_UNDEF)
  35. return "Undefined"sv;
  36. if (index >= SHN_LORESERVE)
  37. return "Reserved"sv;
  38. return section(index).name();
  39. }
  40. unsigned Image::symbol_count() const
  41. {
  42. VERIFY(m_valid);
  43. if (!section_count())
  44. return 0;
  45. return section(m_symbol_table_section_index).entry_count();
  46. }
  47. void Image::dump() const
  48. {
  49. #if ELF_IMAGE_DEBUG
  50. dbgln("ELF::Image({:p}) {{", this);
  51. dbgln(" is_valid: {}", is_valid());
  52. if (!is_valid()) {
  53. dbgln("}}");
  54. return;
  55. }
  56. dbgln(" type: {}", ELF::Image::object_file_type_to_string(header().e_type).value_or("(?)"));
  57. dbgln(" machine: {}", header().e_machine);
  58. dbgln(" entry: {:x}", header().e_entry);
  59. dbgln(" shoff: {}", header().e_shoff);
  60. dbgln(" shnum: {}", header().e_shnum);
  61. dbgln(" phoff: {}", header().e_phoff);
  62. dbgln(" phnum: {}", header().e_phnum);
  63. dbgln(" shstrndx: {}", header().e_shstrndx);
  64. for_each_program_header([&](const ProgramHeader& program_header) {
  65. dbgln(" Program Header {}: {{", program_header.index());
  66. dbgln(" type: {:x}", program_header.type());
  67. dbgln(" offset: {:x}", program_header.offset());
  68. dbgln(" flags: {:x}", program_header.flags());
  69. dbgln(" }}");
  70. });
  71. for (unsigned i = 0; i < header().e_shnum; ++i) {
  72. const auto& section = this->section(i);
  73. dbgln(" Section {}: {{", i);
  74. dbgln(" name: {}", section.name());
  75. dbgln(" type: {:x}", section.type());
  76. dbgln(" offset: {:x}", section.offset());
  77. dbgln(" size: {}", section.size());
  78. dbgln(" ");
  79. dbgln(" }}");
  80. }
  81. dbgln("Symbol count: {} (table is {})", symbol_count(), m_symbol_table_section_index);
  82. for (unsigned i = 1; i < symbol_count(); ++i) {
  83. const auto& sym = symbol(i);
  84. dbgln("Symbol @{}:", i);
  85. dbgln(" Name: {}", sym.name());
  86. dbgln(" In section: {}", section_index_to_string(sym.section_index()));
  87. dbgln(" Value: {}", sym.value());
  88. dbgln(" Size: {}", sym.size());
  89. }
  90. dbgln("}}");
  91. #endif
  92. }
  93. unsigned Image::section_count() const
  94. {
  95. VERIFY(m_valid);
  96. return header().e_shnum;
  97. }
  98. unsigned Image::program_header_count() const
  99. {
  100. VERIFY(m_valid);
  101. return header().e_phnum;
  102. }
  103. bool Image::parse()
  104. {
  105. if (m_size < sizeof(ElfW(Ehdr)) || !validate_elf_header(header(), m_size, m_verbose_logging)) {
  106. if (m_verbose_logging)
  107. dbgln("ELF::Image::parse(): ELF Header not valid");
  108. m_valid = false;
  109. return false;
  110. }
  111. if (!validate_program_headers(header(), m_size, m_buffer, m_size, nullptr, m_verbose_logging)) {
  112. if (m_verbose_logging)
  113. dbgln("ELF::Image::parse(): ELF Program Headers not valid");
  114. m_valid = false;
  115. return false;
  116. }
  117. m_valid = true;
  118. // First locate the string tables.
  119. for (unsigned i = 0; i < section_count(); ++i) {
  120. auto& sh = section_header(i);
  121. if (sh.sh_type == SHT_SYMTAB) {
  122. if (m_symbol_table_section_index && m_symbol_table_section_index != i) {
  123. m_valid = false;
  124. return false;
  125. }
  126. m_symbol_table_section_index = i;
  127. }
  128. if (sh.sh_type == SHT_STRTAB && i != header().e_shstrndx) {
  129. if (section_header_table_string(sh.sh_name) == ELF_STRTAB)
  130. m_string_table_section_index = i;
  131. }
  132. }
  133. return m_valid;
  134. }
  135. StringView Image::table_string(unsigned table_index, unsigned offset) const
  136. {
  137. VERIFY(m_valid);
  138. auto& sh = section_header(table_index);
  139. if (sh.sh_type != SHT_STRTAB)
  140. return nullptr;
  141. size_t computed_offset = sh.sh_offset + offset;
  142. if (computed_offset >= m_size) {
  143. if (m_verbose_logging)
  144. dbgln("SHENANIGANS! Image::table_string() computed offset outside image.");
  145. return {};
  146. }
  147. size_t max_length = min(m_size - computed_offset, (size_t)PAGE_SIZE);
  148. size_t length = strnlen(raw_data(sh.sh_offset + offset), max_length);
  149. return { raw_data(sh.sh_offset + offset), length };
  150. }
  151. StringView Image::section_header_table_string(unsigned offset) const
  152. {
  153. VERIFY(m_valid);
  154. return table_string(header().e_shstrndx, offset);
  155. }
  156. StringView Image::table_string(unsigned offset) const
  157. {
  158. VERIFY(m_valid);
  159. return table_string(m_string_table_section_index, offset);
  160. }
  161. const char* Image::raw_data(unsigned offset) const
  162. {
  163. VERIFY(offset < m_size); // Callers must check indices into raw_data()'s result are also in bounds.
  164. return reinterpret_cast<const char*>(m_buffer) + offset;
  165. }
  166. const ElfW(Ehdr) & Image::header() const
  167. {
  168. VERIFY(m_size >= sizeof(ElfW(Ehdr)));
  169. return *reinterpret_cast<const ElfW(Ehdr)*>(raw_data(0));
  170. }
  171. const ElfW(Phdr) & Image::program_header_internal(unsigned index) const
  172. {
  173. VERIFY(m_valid);
  174. VERIFY(index < header().e_phnum);
  175. return *reinterpret_cast<const ElfW(Phdr)*>(raw_data(header().e_phoff + (index * sizeof(ElfW(Phdr)))));
  176. }
  177. const ElfW(Shdr) & Image::section_header(unsigned index) const
  178. {
  179. VERIFY(m_valid);
  180. VERIFY(index < header().e_shnum);
  181. return *reinterpret_cast<const ElfW(Shdr)*>(raw_data(header().e_shoff + (index * header().e_shentsize)));
  182. }
  183. Image::Symbol Image::symbol(unsigned index) const
  184. {
  185. VERIFY(m_valid);
  186. VERIFY(index < symbol_count());
  187. auto* raw_syms = reinterpret_cast<const ElfW(Sym)*>(raw_data(section(m_symbol_table_section_index).offset()));
  188. return Symbol(*this, index, raw_syms[index]);
  189. }
  190. Image::Section Image::section(unsigned index) const
  191. {
  192. VERIFY(m_valid);
  193. VERIFY(index < section_count());
  194. return Section(*this, index);
  195. }
  196. Image::ProgramHeader Image::program_header(unsigned index) const
  197. {
  198. VERIFY(m_valid);
  199. VERIFY(index < program_header_count());
  200. return ProgramHeader(*this, index);
  201. }
  202. Image::Relocation Image::RelocationSection::relocation(unsigned index) const
  203. {
  204. VERIFY(index < relocation_count());
  205. auto* rels = reinterpret_cast<const ElfW(Rel)*>(m_image.raw_data(offset()));
  206. return Relocation(m_image, rels[index]);
  207. }
  208. Optional<Image::RelocationSection> Image::Section::relocations() const
  209. {
  210. StringBuilder builder;
  211. builder.append(".rel"sv);
  212. builder.append(name());
  213. auto relocation_section = m_image.lookup_section(builder.to_string());
  214. if (!relocation_section.has_value())
  215. return {};
  216. dbgln_if(ELF_IMAGE_DEBUG, "Found relocations for {} in {}", name(), relocation_section.value().name());
  217. return static_cast<RelocationSection>(relocation_section.value());
  218. }
  219. Optional<Image::Section> Image::lookup_section(const StringView& name) const
  220. {
  221. VERIFY(m_valid);
  222. for (unsigned i = 0; i < section_count(); ++i) {
  223. auto section = this->section(i);
  224. if (section.name() == name)
  225. return section;
  226. }
  227. return {};
  228. }
  229. Optional<StringView> Image::object_file_type_to_string(ElfW(Half) type)
  230. {
  231. switch (type) {
  232. case ET_NONE:
  233. return "None"sv;
  234. case ET_REL:
  235. return "Relocatable"sv;
  236. case ET_EXEC:
  237. return "Executable"sv;
  238. case ET_DYN:
  239. return "Shared object"sv;
  240. case ET_CORE:
  241. return "Core"sv;
  242. default:
  243. return {};
  244. }
  245. }
  246. Optional<StringView> Image::object_machine_type_to_string(ElfW(Half) type)
  247. {
  248. switch (type) {
  249. case ET_NONE:
  250. return "None"sv;
  251. case EM_M32:
  252. return "AT&T WE 32100"sv;
  253. case EM_SPARC:
  254. return "SPARC"sv;
  255. case EM_386:
  256. return "Intel 80386"sv;
  257. case EM_68K:
  258. return "Motorola 68000"sv;
  259. case EM_88K:
  260. return "Motorola 88000"sv;
  261. case EM_486:
  262. return "Intel 80486"sv;
  263. case EM_860:
  264. return "Intel 80860"sv;
  265. case EM_MIPS:
  266. return "MIPS R3000 Big-Endian only"sv;
  267. case EM_X86_64:
  268. return "Advanced Micro Devices X86-64"sv;
  269. default:
  270. return {};
  271. }
  272. }
  273. Optional<StringView> Image::object_abi_type_to_string(Elf_Byte type)
  274. {
  275. switch (type) {
  276. case ELFOSABI_SYSV:
  277. return "SYSV"sv;
  278. case ELFOSABI_HPUX:
  279. return "HP-UX"sv;
  280. case ELFOSABI_NETBSD:
  281. return "NetBSD"sv;
  282. case ELFOSABI_LINUX:
  283. return "Linux"sv;
  284. case ELFOSABI_HURD:
  285. return "GNU Hurd"sv;
  286. case ELFOSABI_86OPEN:
  287. return "86Open"sv;
  288. case ELFOSABI_SOLARIS:
  289. return "Solaris"sv;
  290. case ELFOSABI_MONTEREY:
  291. return "AIX"sv;
  292. case ELFOSABI_IRIX:
  293. return "IRIX"sv;
  294. case ELFOSABI_FREEBSD:
  295. return "FreeBSD"sv;
  296. case ELFOSABI_TRU64:
  297. return "Tru64"sv;
  298. case ELFOSABI_MODESTO:
  299. return "Novell Modesto"sv;
  300. case ELFOSABI_OPENBSD:
  301. return "OpenBSD"sv;
  302. case ELFOSABI_ARM:
  303. return "ARM"sv;
  304. case ELFOSABI_STANDALONE:
  305. return "Standalone"sv;
  306. default:
  307. return {};
  308. }
  309. }
  310. StringView Image::Symbol::raw_data() const
  311. {
  312. auto section = this->section();
  313. return { section.raw_data() + (value() - section.address()), size() };
  314. }
  315. #ifndef KERNEL
  316. Optional<Image::Symbol> Image::find_demangled_function(const StringView& name) const
  317. {
  318. Optional<Image::Symbol> found;
  319. for_each_symbol([&](const Image::Symbol& symbol) {
  320. if (symbol.type() != STT_FUNC)
  321. return IterationDecision::Continue;
  322. if (symbol.is_undefined())
  323. return IterationDecision::Continue;
  324. auto demangled = demangle(symbol.name());
  325. auto index_of_paren = demangled.find('(');
  326. if (index_of_paren.has_value()) {
  327. demangled = demangled.substring(0, index_of_paren.value());
  328. }
  329. if (demangled != name)
  330. return IterationDecision::Continue;
  331. found = symbol;
  332. return IterationDecision::Break;
  333. });
  334. return found;
  335. }
  336. #endif
  337. Image::SortedSymbol* Image::find_sorted_symbol(FlatPtr address) const
  338. {
  339. if (m_sorted_symbols.is_empty())
  340. sort_symbols();
  341. size_t index = 0;
  342. binary_search(m_sorted_symbols, nullptr, &index, [&address](auto, auto& candidate) {
  343. if (address < candidate.address)
  344. return -1;
  345. else if (address > candidate.address)
  346. return 1;
  347. else
  348. return 0;
  349. });
  350. // FIXME: The error path here feels strange, index == 0 means error but what about symbol #0?
  351. if (index == 0)
  352. return nullptr;
  353. return &m_sorted_symbols[index];
  354. }
  355. Optional<Image::Symbol> Image::find_symbol(FlatPtr address, u32* out_offset) const
  356. {
  357. auto symbol_count = this->symbol_count();
  358. if (!symbol_count)
  359. return {};
  360. auto* symbol = find_sorted_symbol(address);
  361. if (!symbol)
  362. return {};
  363. if (out_offset)
  364. *out_offset = address - symbol->address;
  365. return symbol->symbol;
  366. }
  367. NEVER_INLINE void Image::sort_symbols() const
  368. {
  369. m_sorted_symbols.ensure_capacity(symbol_count());
  370. for_each_symbol([this](const auto& symbol) {
  371. m_sorted_symbols.append({ symbol.value(), symbol.name(), {}, symbol });
  372. });
  373. quick_sort(m_sorted_symbols, [](auto& a, auto& b) {
  374. return a.address < b.address;
  375. });
  376. }
  377. #ifndef KERNEL
  378. String Image::symbolicate(FlatPtr address, u32* out_offset) const
  379. {
  380. auto symbol_count = this->symbol_count();
  381. if (!symbol_count) {
  382. if (out_offset)
  383. *out_offset = 0;
  384. return "??";
  385. }
  386. auto* symbol = find_sorted_symbol(address);
  387. if (!symbol) {
  388. if (out_offset)
  389. *out_offset = 0;
  390. return "??";
  391. }
  392. auto& demangled_name = symbol->demangled_name;
  393. if (demangled_name.is_null())
  394. demangled_name = demangle(symbol->name);
  395. if (out_offset) {
  396. *out_offset = address - symbol->address;
  397. return demangled_name;
  398. }
  399. return String::formatted("{} +{:#x}", demangled_name, address - symbol->address);
  400. }
  401. #endif
  402. } // end namespace ELF