DynamicLoader.cpp 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559
  1. /*
  2. * Copyright (c) 2019-2020, Andrew Kaster <andrewdkaster@gmail.com>
  3. * Copyright (c) 2020, Itamar S. <itamar8910@gmail.com>
  4. * Copyright (c) 2021, Andreas Kling <kling@serenityos.org>
  5. *
  6. * SPDX-License-Identifier: BSD-2-Clause
  7. */
  8. #include <AK/Debug.h>
  9. #include <AK/Optional.h>
  10. #include <AK/QuickSort.h>
  11. #include <AK/StringBuilder.h>
  12. #include <LibELF/DynamicLinker.h>
  13. #include <LibELF/DynamicLoader.h>
  14. #include <LibELF/Hashes.h>
  15. #include <LibELF/Validation.h>
  16. #include <assert.h>
  17. #include <dlfcn.h>
  18. #include <stdio.h>
  19. #include <stdlib.h>
  20. #include <string.h>
  21. #include <sys/mman.h>
  22. #include <sys/stat.h>
  23. #include <unistd.h>
  24. #ifndef __serenity__
  25. static void* mmap_with_name(void* addr, size_t length, int prot, int flags, int fd, off_t offset, const char*)
  26. {
  27. return mmap(addr, length, prot, flags, fd, offset);
  28. }
  29. # define MAP_RANDOMIZED 0
  30. #endif
  31. namespace ELF {
  32. RefPtr<DynamicLoader> DynamicLoader::try_create(int fd, String filename)
  33. {
  34. struct stat stat;
  35. if (fstat(fd, &stat) < 0) {
  36. perror("DynamicLoader::try_create fstat");
  37. return {};
  38. }
  39. VERIFY(stat.st_size >= 0);
  40. auto size = static_cast<size_t>(stat.st_size);
  41. if (size < sizeof(Elf32_Ehdr))
  42. return {};
  43. String file_mmap_name = String::formatted("ELF_DYN: {}", filename);
  44. auto* data = mmap_with_name(nullptr, size, PROT_READ, MAP_PRIVATE, fd, 0, file_mmap_name.characters());
  45. if (data == MAP_FAILED) {
  46. perror("DynamicLoader::try_create mmap");
  47. return {};
  48. }
  49. return adopt_ref(*new DynamicLoader(fd, move(filename), data, size));
  50. }
  51. DynamicLoader::DynamicLoader(int fd, String filename, void* data, size_t size)
  52. : m_filename(move(filename))
  53. , m_file_size(size)
  54. , m_image_fd(fd)
  55. , m_file_data(data)
  56. , m_elf_image((u8*)m_file_data, m_file_size)
  57. {
  58. m_tls_size = calculate_tls_size();
  59. m_valid = validate();
  60. }
  61. DynamicLoader::~DynamicLoader()
  62. {
  63. if (munmap(m_file_data, m_file_size) < 0) {
  64. perror("munmap");
  65. VERIFY_NOT_REACHED();
  66. }
  67. if (close(m_image_fd) < 0) {
  68. perror("close");
  69. VERIFY_NOT_REACHED();
  70. }
  71. }
  72. const DynamicObject& DynamicLoader::dynamic_object() const
  73. {
  74. if (!m_cached_dynamic_object) {
  75. VirtualAddress dynamic_section_address;
  76. m_elf_image.for_each_program_header([&dynamic_section_address](auto program_header) {
  77. if (program_header.type() == PT_DYNAMIC) {
  78. dynamic_section_address = VirtualAddress(program_header.raw_data());
  79. }
  80. return IterationDecision::Continue;
  81. });
  82. VERIFY(!dynamic_section_address.is_null());
  83. m_cached_dynamic_object = ELF::DynamicObject::create(m_filename, VirtualAddress(m_elf_image.base_address()), dynamic_section_address);
  84. }
  85. return *m_cached_dynamic_object;
  86. }
  87. size_t DynamicLoader::calculate_tls_size() const
  88. {
  89. size_t tls_size = 0;
  90. m_elf_image.for_each_program_header([&tls_size](auto program_header) {
  91. if (program_header.type() == PT_TLS) {
  92. tls_size = program_header.size_in_memory();
  93. }
  94. return IterationDecision::Continue;
  95. });
  96. return tls_size;
  97. }
  98. bool DynamicLoader::validate()
  99. {
  100. if (!m_elf_image.is_valid())
  101. return false;
  102. auto* elf_header = (Elf32_Ehdr*)m_file_data;
  103. if (!validate_elf_header(*elf_header, m_file_size))
  104. return false;
  105. if (!validate_program_headers(*elf_header, m_file_size, (u8*)m_file_data, m_file_size, &m_program_interpreter))
  106. return false;
  107. return true;
  108. }
  109. void* DynamicLoader::symbol_for_name(const StringView& name)
  110. {
  111. auto result = m_dynamic_object->hash_section().lookup_symbol(name);
  112. if (!result.has_value())
  113. return nullptr;
  114. auto symbol = result.value();
  115. if (symbol.is_undefined())
  116. return nullptr;
  117. return m_dynamic_object->base_address().offset(symbol.value()).as_ptr();
  118. }
  119. RefPtr<DynamicObject> DynamicLoader::map()
  120. {
  121. if (m_dynamic_object) {
  122. // Already mapped.
  123. return nullptr;
  124. }
  125. if (!m_valid) {
  126. dbgln("DynamicLoader::map failed: image is invalid");
  127. return nullptr;
  128. }
  129. load_program_headers();
  130. VERIFY(!m_base_address.is_null());
  131. m_dynamic_object = DynamicObject::create(m_filename, m_base_address, m_dynamic_section_address);
  132. m_dynamic_object->set_tls_offset(m_tls_offset);
  133. m_dynamic_object->set_tls_size(m_tls_size);
  134. return m_dynamic_object;
  135. }
  136. bool DynamicLoader::link(unsigned flags, size_t total_tls_size)
  137. {
  138. return load_stage_2(flags, total_tls_size);
  139. }
  140. bool DynamicLoader::load_stage_2(unsigned flags, size_t total_tls_size)
  141. {
  142. VERIFY(flags & RTLD_GLOBAL);
  143. if (m_dynamic_object->has_text_relocations()) {
  144. for (auto& text_segment : m_text_segments) {
  145. VERIFY(text_segment.address().get() != 0);
  146. #ifndef AK_OS_MACOS
  147. // Remap this text region as private.
  148. if (mremap(text_segment.address().as_ptr(), text_segment.size(), text_segment.size(), MAP_PRIVATE) == MAP_FAILED) {
  149. perror("mremap .text: MAP_PRIVATE");
  150. return false;
  151. }
  152. #endif
  153. if (0 > mprotect(text_segment.address().as_ptr(), text_segment.size(), PROT_READ | PROT_WRITE)) {
  154. perror("mprotect .text: PROT_READ | PROT_WRITE"); // FIXME: dlerror?
  155. return false;
  156. }
  157. }
  158. }
  159. do_main_relocations(total_tls_size);
  160. return true;
  161. }
  162. void DynamicLoader::do_main_relocations(size_t total_tls_size)
  163. {
  164. auto do_single_relocation = [&](const ELF::DynamicObject::Relocation& relocation) {
  165. switch (do_relocation(total_tls_size, relocation, ShouldInitializeWeak::No)) {
  166. case RelocationResult::Failed:
  167. dbgln("Loader.so: {} unresolved symbol '{}'", m_filename, relocation.symbol().name());
  168. VERIFY_NOT_REACHED();
  169. case RelocationResult::ResolveLater:
  170. m_unresolved_relocations.append(relocation);
  171. break;
  172. case RelocationResult::Success:
  173. break;
  174. }
  175. return IterationDecision::Continue;
  176. };
  177. m_dynamic_object->relocation_section().for_each_relocation(do_single_relocation);
  178. m_dynamic_object->plt_relocation_section().for_each_relocation(do_single_relocation);
  179. }
  180. RefPtr<DynamicObject> DynamicLoader::load_stage_3(unsigned flags, size_t total_tls_size)
  181. {
  182. do_lazy_relocations(total_tls_size);
  183. if (flags & RTLD_LAZY) {
  184. if (m_dynamic_object->has_plt())
  185. setup_plt_trampoline();
  186. }
  187. for (auto& text_segment : m_text_segments) {
  188. if (mprotect(text_segment.address().as_ptr(), text_segment.size(), PROT_READ | PROT_EXEC) < 0) {
  189. perror("mprotect .text: PROT_READ | PROT_EXEC"); // FIXME: dlerror?
  190. return nullptr;
  191. }
  192. }
  193. if (m_relro_segment_size) {
  194. if (mprotect(m_relro_segment_address.as_ptr(), m_relro_segment_size, PROT_READ) < 0) {
  195. perror("mprotect .relro: PROT_READ");
  196. return nullptr;
  197. }
  198. #if __serenity__
  199. if (set_mmap_name(m_relro_segment_address.as_ptr(), m_relro_segment_size, String::formatted("{}: .relro", m_filename).characters()) < 0) {
  200. perror("set_mmap_name .relro");
  201. return nullptr;
  202. }
  203. #endif
  204. }
  205. return m_dynamic_object;
  206. }
  207. void DynamicLoader::load_stage_4()
  208. {
  209. call_object_init_functions();
  210. }
  211. void DynamicLoader::do_lazy_relocations(size_t total_tls_size)
  212. {
  213. for (const auto& relocation : m_unresolved_relocations) {
  214. if (auto res = do_relocation(total_tls_size, relocation, ShouldInitializeWeak::Yes); res != RelocationResult::Success) {
  215. dbgln("Loader.so: {} unresolved symbol '{}'", m_filename, relocation.symbol().name());
  216. VERIFY_NOT_REACHED();
  217. }
  218. }
  219. }
  220. void DynamicLoader::load_program_headers()
  221. {
  222. Vector<ProgramHeaderRegion> load_regions;
  223. Vector<ProgramHeaderRegion> text_regions;
  224. Vector<ProgramHeaderRegion> data_regions;
  225. Optional<ProgramHeaderRegion> tls_region;
  226. Optional<ProgramHeaderRegion> relro_region;
  227. VirtualAddress dynamic_region_desired_vaddr;
  228. m_elf_image.for_each_program_header([&](const Image::ProgramHeader& program_header) {
  229. ProgramHeaderRegion region {};
  230. region.set_program_header(program_header.raw_header());
  231. if (region.is_tls_template()) {
  232. VERIFY(!tls_region.has_value());
  233. tls_region = region;
  234. } else if (region.is_load()) {
  235. load_regions.append(region);
  236. if (region.is_executable()) {
  237. text_regions.append(region);
  238. } else {
  239. data_regions.append(region);
  240. }
  241. } else if (region.is_dynamic()) {
  242. dynamic_region_desired_vaddr = region.desired_load_address();
  243. } else if (region.is_relro()) {
  244. VERIFY(!relro_region.has_value());
  245. relro_region = region;
  246. }
  247. return IterationDecision::Continue;
  248. });
  249. VERIFY(!text_regions.is_empty() || !data_regions.is_empty());
  250. auto compare_load_address = [](ProgramHeaderRegion& a, ProgramHeaderRegion& b) {
  251. return a.desired_load_address().as_ptr() < b.desired_load_address().as_ptr();
  252. };
  253. quick_sort(load_regions, compare_load_address);
  254. quick_sort(text_regions, compare_load_address);
  255. quick_sort(data_regions, compare_load_address);
  256. // Process regions in order: .text, .data, .tls
  257. void* requested_load_address = m_elf_image.is_dynamic() ? nullptr : load_regions.first().desired_load_address().as_ptr();
  258. int reservation_mmap_flags = MAP_ANON | MAP_PRIVATE | MAP_NORESERVE;
  259. if (m_elf_image.is_dynamic())
  260. reservation_mmap_flags |= MAP_RANDOMIZED;
  261. else
  262. reservation_mmap_flags |= MAP_FIXED;
  263. for (auto& text_region : text_regions)
  264. VERIFY(!text_region.is_writable());
  265. // First, we make a dummy reservation mapping, in order to allocate enough VM
  266. // to hold all regions contiguously in the address space.
  267. FlatPtr ph_load_base = load_regions.first().desired_load_address().page_base().get();
  268. FlatPtr ph_load_end = round_up_to_power_of_two(load_regions.last().desired_load_address().offset(load_regions.last().size_in_memory()).get(), PAGE_SIZE);
  269. size_t total_mapping_size = ph_load_end - ph_load_base;
  270. auto* reservation = mmap(requested_load_address, total_mapping_size, PROT_NONE, reservation_mmap_flags, 0, 0);
  271. if (reservation == MAP_FAILED) {
  272. perror("mmap reservation");
  273. VERIFY_NOT_REACHED();
  274. }
  275. m_base_address = VirtualAddress { reservation };
  276. // Then we unmap the reservation.
  277. if (munmap(reservation, total_mapping_size) < 0) {
  278. perror("munmap reservation");
  279. VERIFY_NOT_REACHED();
  280. }
  281. for (auto& text_region : text_regions) {
  282. FlatPtr ph_text_base = text_region.desired_load_address().page_base().get();
  283. FlatPtr ph_text_end = round_up_to_power_of_two(text_region.desired_load_address().offset(text_region.size_in_memory()).get(), PAGE_SIZE);
  284. size_t text_segment_size = ph_text_end - ph_text_base;
  285. auto text_segment_offset = ph_text_base - ph_load_base;
  286. auto* text_segment_address = (u8*)reservation + text_segment_offset;
  287. // Now we can map the text segment at the reserved address.
  288. auto* text_segment_begin = (u8*)mmap_with_name(
  289. text_segment_address,
  290. text_segment_size,
  291. PROT_READ,
  292. MAP_FILE | MAP_SHARED | MAP_FIXED,
  293. m_image_fd,
  294. text_region.offset(),
  295. String::formatted("{}: .text", m_filename).characters());
  296. if (text_segment_begin == MAP_FAILED) {
  297. perror("mmap text");
  298. VERIFY_NOT_REACHED();
  299. }
  300. m_text_segments.append({ VirtualAddress { (FlatPtr)text_segment_begin }, text_segment_size });
  301. }
  302. VERIFY(requested_load_address == nullptr || requested_load_address == reservation);
  303. if (relro_region.has_value()) {
  304. m_relro_segment_size = relro_region->size_in_memory();
  305. m_relro_segment_address = VirtualAddress { (u8*)reservation + relro_region->desired_load_address().get() };
  306. }
  307. if (m_elf_image.is_dynamic())
  308. m_dynamic_section_address = VirtualAddress { (u8*)reservation + dynamic_region_desired_vaddr.get() };
  309. else
  310. m_dynamic_section_address = dynamic_region_desired_vaddr;
  311. for (auto& data_region : data_regions) {
  312. FlatPtr ph_data_base = data_region.desired_load_address().page_base().get();
  313. FlatPtr ph_data_end = round_up_to_power_of_two(data_region.desired_load_address().offset(data_region.size_in_memory()).get(), PAGE_SIZE);
  314. size_t data_segment_size = ph_data_end - ph_data_base;
  315. auto data_segment_offset = ph_data_base - ph_load_base;
  316. auto* data_segment_address = (u8*)reservation + data_segment_offset;
  317. // Finally, we make an anonymous mapping for the data segment. Contents are then copied from the file.
  318. auto* data_segment = (u8*)mmap_with_name(
  319. data_segment_address,
  320. data_segment_size,
  321. PROT_READ | PROT_WRITE,
  322. MAP_ANONYMOUS | MAP_PRIVATE | MAP_FIXED,
  323. 0,
  324. 0,
  325. String::formatted("{}: .data", m_filename).characters());
  326. if (MAP_FAILED == data_segment) {
  327. perror("mmap data");
  328. VERIFY_NOT_REACHED();
  329. }
  330. VirtualAddress data_segment_start;
  331. if (m_elf_image.is_dynamic())
  332. data_segment_start = VirtualAddress { (u8*)reservation + data_region.desired_load_address().get() };
  333. else
  334. data_segment_start = data_region.desired_load_address();
  335. memcpy(data_segment_start.as_ptr(), (u8*)m_file_data + data_region.offset(), data_region.size_in_image());
  336. }
  337. // FIXME: Initialize the values in the TLS section. Currently, it is zeroed.
  338. }
  339. DynamicLoader::RelocationResult DynamicLoader::do_relocation(size_t total_tls_size, const ELF::DynamicObject::Relocation& relocation, ShouldInitializeWeak should_initialize_weak)
  340. {
  341. FlatPtr* patch_ptr = nullptr;
  342. if (is_dynamic())
  343. patch_ptr = (FlatPtr*)(m_dynamic_object->base_address().as_ptr() + relocation.offset());
  344. else
  345. patch_ptr = (FlatPtr*)(FlatPtr)relocation.offset();
  346. switch (relocation.type()) {
  347. case R_386_NONE:
  348. // Apparently most loaders will just skip these?
  349. // Seems if the 'link editor' generates one something is funky with your code
  350. break;
  351. case R_386_32: {
  352. auto symbol = relocation.symbol();
  353. auto res = lookup_symbol(symbol);
  354. if (!res.has_value()) {
  355. if (symbol.bind() == STB_WEAK)
  356. return RelocationResult::ResolveLater;
  357. dbgln("ERROR: symbol not found: {}.", symbol.name());
  358. return RelocationResult::Failed;
  359. }
  360. auto symbol_address = res.value().address;
  361. *patch_ptr += symbol_address.get();
  362. break;
  363. }
  364. case R_386_PC32: {
  365. auto symbol = relocation.symbol();
  366. auto result = lookup_symbol(symbol);
  367. if (!result.has_value())
  368. return RelocationResult::Failed;
  369. auto relative_offset = result.value().address - m_dynamic_object->base_address().offset(relocation.offset());
  370. *patch_ptr += relative_offset.get();
  371. break;
  372. }
  373. case R_386_GLOB_DAT: {
  374. auto symbol = relocation.symbol();
  375. auto res = lookup_symbol(symbol);
  376. VirtualAddress symbol_location;
  377. if (!res.has_value()) {
  378. if (symbol.bind() == STB_WEAK) {
  379. if (should_initialize_weak == ShouldInitializeWeak::No)
  380. return RelocationResult::ResolveLater;
  381. } else {
  382. // Symbol not found
  383. return RelocationResult::Failed;
  384. }
  385. symbol_location = VirtualAddress { (FlatPtr)0 };
  386. } else
  387. symbol_location = res.value().address;
  388. VERIFY(symbol_location != m_dynamic_object->base_address());
  389. *patch_ptr = symbol_location.get();
  390. break;
  391. }
  392. case R_386_RELATIVE: {
  393. // FIXME: According to the spec, R_386_relative ones must be done first.
  394. // We could explicitly do them first using m_number_of_relocations from DT_RELCOUNT
  395. // However, our compiler is nice enough to put them at the front of the relocations for us :)
  396. *patch_ptr += (FlatPtr)m_dynamic_object->base_address().as_ptr(); // + addend for RelA (addend for Rel is stored at addr)
  397. break;
  398. }
  399. case R_386_TLS_TPOFF32:
  400. case R_386_TLS_TPOFF: {
  401. auto symbol = relocation.symbol();
  402. // For some reason, LibC has a R_386_TLS_TPOFF that refers to the undefined symbol.. huh
  403. if (relocation.symbol_index() == 0)
  404. break;
  405. auto res = lookup_symbol(symbol);
  406. if (!res.has_value())
  407. break;
  408. u32 symbol_value = res.value().value;
  409. auto* dynamic_object_of_symbol = res.value().dynamic_object;
  410. VERIFY(dynamic_object_of_symbol);
  411. *patch_ptr = dynamic_object_of_symbol->tls_offset().value() + symbol_value - total_tls_size;
  412. break;
  413. }
  414. case R_386_JMP_SLOT: {
  415. // FIXME: Or BIND_NOW flag passed in?
  416. if (m_dynamic_object->must_bind_now()) {
  417. // Eagerly BIND_NOW the PLT entries, doing all the symbol looking goodness
  418. // The patch method returns the address for the LAZY fixup path, but we don't need it here
  419. m_dynamic_object->patch_plt_entry(relocation.offset_in_section());
  420. } else {
  421. u8* relocation_address = relocation.address().as_ptr();
  422. if (m_elf_image.is_dynamic())
  423. *(u32*)relocation_address += (FlatPtr)m_dynamic_object->base_address().as_ptr();
  424. }
  425. break;
  426. }
  427. default:
  428. // Raise the alarm! Someone needs to implement this relocation type
  429. dbgln("Found a new exciting relocation type {}", relocation.type());
  430. VERIFY_NOT_REACHED();
  431. }
  432. return RelocationResult::Success;
  433. }
  434. // Defined in <arch>/plt_trampoline.S
  435. extern "C" void _plt_trampoline(void) __attribute__((visibility("hidden")));
  436. void DynamicLoader::setup_plt_trampoline()
  437. {
  438. VERIFY(m_dynamic_object);
  439. VERIFY(m_dynamic_object->has_plt());
  440. VirtualAddress got_address = m_dynamic_object->plt_got_base_address();
  441. auto* got_ptr = (FlatPtr*)got_address.as_ptr();
  442. got_ptr[1] = (FlatPtr)m_dynamic_object.ptr();
  443. got_ptr[2] = (FlatPtr)&_plt_trampoline;
  444. }
  445. // Called from our ASM routine _plt_trampoline.
  446. // Tell the compiler that it might be called from other places:
  447. extern "C" FlatPtr _fixup_plt_entry(DynamicObject* object, u32 relocation_offset);
  448. extern "C" FlatPtr _fixup_plt_entry(DynamicObject* object, u32 relocation_offset)
  449. {
  450. return object->patch_plt_entry(relocation_offset).get();
  451. }
  452. void DynamicLoader::call_object_init_functions()
  453. {
  454. typedef void (*InitFunc)();
  455. if (m_dynamic_object->has_init_section()) {
  456. auto init_function = (InitFunc)(m_dynamic_object->init_section().address().as_ptr());
  457. (init_function)();
  458. }
  459. if (m_dynamic_object->has_init_array_section()) {
  460. auto init_array_section = m_dynamic_object->init_array_section();
  461. InitFunc* init_begin = (InitFunc*)(init_array_section.address().as_ptr());
  462. InitFunc* init_end = init_begin + init_array_section.entry_count();
  463. while (init_begin != init_end) {
  464. // Android sources claim that these can be -1, to be ignored.
  465. // 0 definitely shows up. Apparently 0/-1 are valid? Confusing.
  466. if (!*init_begin || ((FlatPtr)*init_begin == (FlatPtr)-1))
  467. continue;
  468. (*init_begin)();
  469. ++init_begin;
  470. }
  471. }
  472. }
  473. Optional<DynamicObject::SymbolLookupResult> DynamicLoader::lookup_symbol(const ELF::DynamicObject::Symbol& symbol)
  474. {
  475. if (symbol.is_undefined() || symbol.bind() == STB_WEAK)
  476. return DynamicLinker::lookup_global_symbol(symbol.name());
  477. return DynamicObject::SymbolLookupResult { symbol.value(), symbol.address(), symbol.bind(), &symbol.object() };
  478. }
  479. } // end namespace ELF