DynamicLoader.cpp 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552
  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 <LibDl/dlfcn.h>
  13. #include <LibDl/dlfcn_integration.h>
  14. #include <LibELF/DynamicLinker.h>
  15. #include <LibELF/DynamicLoader.h>
  16. #include <LibELF/Hashes.h>
  17. #include <LibELF/Validation.h>
  18. #include <assert.h>
  19. #include <errno.h>
  20. #include <stdio.h>
  21. #include <stdlib.h>
  22. #include <string.h>
  23. #include <sys/mman.h>
  24. #include <sys/stat.h>
  25. #include <unistd.h>
  26. #ifndef __serenity__
  27. static void* mmap_with_name(void* addr, size_t length, int prot, int flags, int fd, off_t offset, const char*)
  28. {
  29. return mmap(addr, length, prot, flags, fd, offset);
  30. }
  31. # define MAP_RANDOMIZED 0
  32. #endif
  33. namespace ELF {
  34. Result<NonnullRefPtr<DynamicLoader>, DlErrorMessage> DynamicLoader::try_create(int fd, String filename)
  35. {
  36. struct stat stat;
  37. if (fstat(fd, &stat) < 0) {
  38. return DlErrorMessage { "DynamicLoader::try_create fstat" };
  39. }
  40. VERIFY(stat.st_size >= 0);
  41. auto size = static_cast<size_t>(stat.st_size);
  42. if (size < sizeof(Elf32_Ehdr))
  43. return DlErrorMessage { String::formatted("File {} has invalid ELF header", filename) };
  44. String file_mmap_name = String::formatted("ELF_DYN: {}", filename);
  45. auto* data = mmap_with_name(nullptr, size, PROT_READ, MAP_PRIVATE, fd, 0, file_mmap_name.characters());
  46. if (data == MAP_FAILED) {
  47. return DlErrorMessage { "DynamicLoader::try_create mmap" };
  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_of_current_object = 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. RefPtr<DynamicObject> DynamicLoader::map()
  110. {
  111. if (m_dynamic_object) {
  112. // Already mapped.
  113. return nullptr;
  114. }
  115. if (!m_valid) {
  116. dbgln("DynamicLoader::map failed: image is invalid");
  117. return nullptr;
  118. }
  119. load_program_headers();
  120. VERIFY(!m_base_address.is_null());
  121. m_dynamic_object = DynamicObject::create(m_filename, m_base_address, m_dynamic_section_address);
  122. m_dynamic_object->set_tls_offset(m_tls_offset);
  123. m_dynamic_object->set_tls_size(m_tls_size_of_current_object);
  124. return m_dynamic_object;
  125. }
  126. bool DynamicLoader::link(unsigned flags, size_t total_tls_size)
  127. {
  128. return load_stage_2(flags, total_tls_size);
  129. }
  130. bool DynamicLoader::load_stage_2(unsigned flags, size_t total_tls_size)
  131. {
  132. VERIFY(flags & RTLD_GLOBAL);
  133. if (m_dynamic_object->has_text_relocations()) {
  134. for (auto& text_segment : m_text_segments) {
  135. VERIFY(text_segment.address().get() != 0);
  136. #ifndef AK_OS_MACOS
  137. // Remap this text region as private.
  138. if (mremap(text_segment.address().as_ptr(), text_segment.size(), text_segment.size(), MAP_PRIVATE) == MAP_FAILED) {
  139. perror("mremap .text: MAP_PRIVATE");
  140. return false;
  141. }
  142. #endif
  143. if (0 > mprotect(text_segment.address().as_ptr(), text_segment.size(), PROT_READ | PROT_WRITE)) {
  144. perror("mprotect .text: PROT_READ | PROT_WRITE"); // FIXME: dlerror?
  145. return false;
  146. }
  147. }
  148. }
  149. do_main_relocations(total_tls_size);
  150. return true;
  151. }
  152. void DynamicLoader::do_main_relocations(size_t total_tls_size)
  153. {
  154. auto do_single_relocation = [&](const ELF::DynamicObject::Relocation& relocation) {
  155. switch (do_relocation(total_tls_size, relocation, ShouldInitializeWeak::No)) {
  156. case RelocationResult::Failed:
  157. dbgln("Loader.so: {} unresolved symbol '{}'", m_filename, relocation.symbol().name());
  158. VERIFY_NOT_REACHED();
  159. case RelocationResult::ResolveLater:
  160. m_unresolved_relocations.append(relocation);
  161. break;
  162. case RelocationResult::Success:
  163. break;
  164. }
  165. return IterationDecision::Continue;
  166. };
  167. m_dynamic_object->relocation_section().for_each_relocation(do_single_relocation);
  168. m_dynamic_object->plt_relocation_section().for_each_relocation(do_single_relocation);
  169. }
  170. Result<NonnullRefPtr<DynamicObject>, DlErrorMessage> DynamicLoader::load_stage_3(unsigned flags, size_t total_tls_size)
  171. {
  172. do_lazy_relocations(total_tls_size);
  173. if (flags & RTLD_LAZY) {
  174. if (m_dynamic_object->has_plt())
  175. setup_plt_trampoline();
  176. }
  177. for (auto& text_segment : m_text_segments) {
  178. if (mprotect(text_segment.address().as_ptr(), text_segment.size(), PROT_READ | PROT_EXEC) < 0) {
  179. return DlErrorMessage { String::formatted("mprotect .text: PROT_READ | PROT_EXEC: {}", strerror(errno)) };
  180. }
  181. }
  182. if (m_relro_segment_size) {
  183. if (mprotect(m_relro_segment_address.as_ptr(), m_relro_segment_size, PROT_READ) < 0) {
  184. return DlErrorMessage { String::formatted("mprotect .text: PROT_READ: {}", strerror(errno)) };
  185. }
  186. #if __serenity__
  187. if (set_mmap_name(m_relro_segment_address.as_ptr(), m_relro_segment_size, String::formatted("{}: .relro", m_filename).characters()) < 0) {
  188. return DlErrorMessage { String::formatted("set_mmap_name .relro: {}", strerror(errno)) };
  189. }
  190. #endif
  191. }
  192. return NonnullRefPtr<DynamicObject> { *m_dynamic_object };
  193. }
  194. void DynamicLoader::load_stage_4()
  195. {
  196. call_object_init_functions();
  197. }
  198. void DynamicLoader::do_lazy_relocations(size_t total_tls_size)
  199. {
  200. for (const auto& relocation : m_unresolved_relocations) {
  201. if (auto res = do_relocation(total_tls_size, relocation, ShouldInitializeWeak::Yes); res != RelocationResult::Success) {
  202. dbgln("Loader.so: {} unresolved symbol '{}'", m_filename, relocation.symbol().name());
  203. VERIFY_NOT_REACHED();
  204. }
  205. }
  206. }
  207. void DynamicLoader::load_program_headers()
  208. {
  209. Vector<ProgramHeaderRegion> load_regions;
  210. Vector<ProgramHeaderRegion> text_regions;
  211. Vector<ProgramHeaderRegion> data_regions;
  212. Optional<ProgramHeaderRegion> tls_region;
  213. Optional<ProgramHeaderRegion> relro_region;
  214. VirtualAddress dynamic_region_desired_vaddr;
  215. m_elf_image.for_each_program_header([&](const Image::ProgramHeader& program_header) {
  216. ProgramHeaderRegion region {};
  217. region.set_program_header(program_header.raw_header());
  218. if (region.is_tls_template()) {
  219. VERIFY(!tls_region.has_value());
  220. tls_region = region;
  221. } else if (region.is_load()) {
  222. load_regions.append(region);
  223. if (region.is_executable()) {
  224. text_regions.append(region);
  225. } else {
  226. data_regions.append(region);
  227. }
  228. } else if (region.is_dynamic()) {
  229. dynamic_region_desired_vaddr = region.desired_load_address();
  230. } else if (region.is_relro()) {
  231. VERIFY(!relro_region.has_value());
  232. relro_region = region;
  233. }
  234. return IterationDecision::Continue;
  235. });
  236. VERIFY(!text_regions.is_empty() || !data_regions.is_empty());
  237. auto compare_load_address = [](ProgramHeaderRegion& a, ProgramHeaderRegion& b) {
  238. return a.desired_load_address().as_ptr() < b.desired_load_address().as_ptr();
  239. };
  240. quick_sort(load_regions, compare_load_address);
  241. quick_sort(text_regions, compare_load_address);
  242. quick_sort(data_regions, compare_load_address);
  243. // Process regions in order: .text, .data, .tls
  244. void* requested_load_address = m_elf_image.is_dynamic() ? nullptr : load_regions.first().desired_load_address().as_ptr();
  245. int reservation_mmap_flags = MAP_ANON | MAP_PRIVATE | MAP_NORESERVE;
  246. if (m_elf_image.is_dynamic())
  247. reservation_mmap_flags |= MAP_RANDOMIZED;
  248. else
  249. reservation_mmap_flags |= MAP_FIXED;
  250. for (auto& text_region : text_regions)
  251. VERIFY(!text_region.is_writable());
  252. // First, we make a dummy reservation mapping, in order to allocate enough VM
  253. // to hold all regions contiguously in the address space.
  254. FlatPtr ph_load_base = load_regions.first().desired_load_address().page_base().get();
  255. 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);
  256. size_t total_mapping_size = ph_load_end - ph_load_base;
  257. auto* reservation = mmap(requested_load_address, total_mapping_size, PROT_NONE, reservation_mmap_flags, 0, 0);
  258. if (reservation == MAP_FAILED) {
  259. perror("mmap reservation");
  260. VERIFY_NOT_REACHED();
  261. }
  262. m_base_address = VirtualAddress { reservation };
  263. // Then we unmap the reservation.
  264. if (munmap(reservation, total_mapping_size) < 0) {
  265. perror("munmap reservation");
  266. VERIFY_NOT_REACHED();
  267. }
  268. for (auto& text_region : text_regions) {
  269. FlatPtr ph_text_base = text_region.desired_load_address().page_base().get();
  270. FlatPtr ph_text_end = round_up_to_power_of_two(text_region.desired_load_address().offset(text_region.size_in_memory()).get(), PAGE_SIZE);
  271. size_t text_segment_size = ph_text_end - ph_text_base;
  272. auto text_segment_offset = ph_text_base - ph_load_base;
  273. auto* text_segment_address = (u8*)reservation + text_segment_offset;
  274. // Now we can map the text segment at the reserved address.
  275. auto* text_segment_begin = (u8*)mmap_with_name(
  276. text_segment_address,
  277. text_segment_size,
  278. PROT_READ,
  279. MAP_FILE | MAP_SHARED | MAP_FIXED,
  280. m_image_fd,
  281. text_region.offset(),
  282. String::formatted("{}: .text", m_filename).characters());
  283. if (text_segment_begin == MAP_FAILED) {
  284. perror("mmap text");
  285. VERIFY_NOT_REACHED();
  286. }
  287. m_text_segments.append({ VirtualAddress { (FlatPtr)text_segment_begin }, text_segment_size });
  288. }
  289. VERIFY(requested_load_address == nullptr || requested_load_address == reservation);
  290. if (relro_region.has_value()) {
  291. m_relro_segment_size = relro_region->size_in_memory();
  292. m_relro_segment_address = VirtualAddress { (u8*)reservation + relro_region->desired_load_address().get() };
  293. }
  294. if (m_elf_image.is_dynamic())
  295. m_dynamic_section_address = VirtualAddress { (u8*)reservation + dynamic_region_desired_vaddr.get() };
  296. else
  297. m_dynamic_section_address = dynamic_region_desired_vaddr;
  298. for (auto& data_region : data_regions) {
  299. FlatPtr ph_data_base = data_region.desired_load_address().page_base().get();
  300. FlatPtr ph_data_end = round_up_to_power_of_two(data_region.desired_load_address().offset(data_region.size_in_memory()).get(), PAGE_SIZE);
  301. size_t data_segment_size = ph_data_end - ph_data_base;
  302. auto data_segment_offset = ph_data_base - ph_load_base;
  303. auto* data_segment_address = (u8*)reservation + data_segment_offset;
  304. // Finally, we make an anonymous mapping for the data segment. Contents are then copied from the file.
  305. auto* data_segment = (u8*)mmap_with_name(
  306. data_segment_address,
  307. data_segment_size,
  308. PROT_READ | PROT_WRITE,
  309. MAP_ANONYMOUS | MAP_PRIVATE | MAP_FIXED,
  310. 0,
  311. 0,
  312. String::formatted("{}: .data", m_filename).characters());
  313. if (MAP_FAILED == data_segment) {
  314. perror("mmap data");
  315. VERIFY_NOT_REACHED();
  316. }
  317. VirtualAddress data_segment_start;
  318. if (m_elf_image.is_dynamic())
  319. data_segment_start = VirtualAddress { (u8*)reservation + data_region.desired_load_address().get() };
  320. else
  321. data_segment_start = data_region.desired_load_address();
  322. memcpy(data_segment_start.as_ptr(), (u8*)m_file_data + data_region.offset(), data_region.size_in_image());
  323. }
  324. // FIXME: Initialize the values in the TLS section. Currently, it is zeroed.
  325. }
  326. DynamicLoader::RelocationResult DynamicLoader::do_relocation(size_t total_tls_size, const ELF::DynamicObject::Relocation& relocation, ShouldInitializeWeak should_initialize_weak)
  327. {
  328. FlatPtr* patch_ptr = nullptr;
  329. if (is_dynamic())
  330. patch_ptr = (FlatPtr*)(m_dynamic_object->base_address().as_ptr() + relocation.offset());
  331. else
  332. patch_ptr = (FlatPtr*)(FlatPtr)relocation.offset();
  333. switch (relocation.type()) {
  334. case R_386_NONE:
  335. // Apparently most loaders will just skip these?
  336. // Seems if the 'link editor' generates one something is funky with your code
  337. break;
  338. case R_386_32: {
  339. auto symbol = relocation.symbol();
  340. auto res = lookup_symbol(symbol);
  341. if (!res.has_value()) {
  342. if (symbol.bind() == STB_WEAK)
  343. return RelocationResult::ResolveLater;
  344. dbgln("ERROR: symbol not found: {}.", symbol.name());
  345. return RelocationResult::Failed;
  346. }
  347. auto symbol_address = res.value().address;
  348. *patch_ptr += symbol_address.get();
  349. break;
  350. }
  351. case R_386_PC32: {
  352. auto symbol = relocation.symbol();
  353. auto result = lookup_symbol(symbol);
  354. if (!result.has_value())
  355. return RelocationResult::Failed;
  356. auto relative_offset = result.value().address - m_dynamic_object->base_address().offset(relocation.offset());
  357. *patch_ptr += relative_offset.get();
  358. break;
  359. }
  360. case R_386_GLOB_DAT: {
  361. auto symbol = relocation.symbol();
  362. auto res = lookup_symbol(symbol);
  363. VirtualAddress symbol_location;
  364. if (!res.has_value()) {
  365. if (symbol.bind() == STB_WEAK) {
  366. if (should_initialize_weak == ShouldInitializeWeak::No)
  367. return RelocationResult::ResolveLater;
  368. } else {
  369. // Symbol not found
  370. return RelocationResult::Failed;
  371. }
  372. symbol_location = VirtualAddress { (FlatPtr)0 };
  373. } else
  374. symbol_location = res.value().address;
  375. VERIFY(symbol_location != m_dynamic_object->base_address());
  376. *patch_ptr = symbol_location.get();
  377. break;
  378. }
  379. case R_386_RELATIVE: {
  380. // FIXME: According to the spec, R_386_relative ones must be done first.
  381. // We could explicitly do them first using m_number_of_relocations from DT_RELCOUNT
  382. // However, our compiler is nice enough to put them at the front of the relocations for us :)
  383. *patch_ptr += (FlatPtr)m_dynamic_object->base_address().as_ptr(); // + addend for RelA (addend for Rel is stored at addr)
  384. break;
  385. }
  386. case R_386_TLS_TPOFF32:
  387. case R_386_TLS_TPOFF: {
  388. auto symbol = relocation.symbol();
  389. // For some reason, LibC has a R_386_TLS_TPOFF that refers to the undefined symbol.. huh
  390. if (relocation.symbol_index() == 0)
  391. break;
  392. auto res = lookup_symbol(symbol);
  393. if (!res.has_value())
  394. break;
  395. auto* dynamic_object_of_symbol = res.value().dynamic_object;
  396. VERIFY(dynamic_object_of_symbol);
  397. *patch_ptr = negative_offset_from_tls_block_end(res.value().value, dynamic_object_of_symbol->tls_offset().value(), total_tls_size);
  398. break;
  399. }
  400. case R_386_JMP_SLOT: {
  401. // FIXME: Or BIND_NOW flag passed in?
  402. if (m_dynamic_object->must_bind_now()) {
  403. // Eagerly BIND_NOW the PLT entries, doing all the symbol looking goodness
  404. // The patch method returns the address for the LAZY fixup path, but we don't need it here
  405. m_dynamic_object->patch_plt_entry(relocation.offset_in_section());
  406. } else {
  407. u8* relocation_address = relocation.address().as_ptr();
  408. if (m_elf_image.is_dynamic())
  409. *(u32*)relocation_address += (FlatPtr)m_dynamic_object->base_address().as_ptr();
  410. }
  411. break;
  412. }
  413. default:
  414. // Raise the alarm! Someone needs to implement this relocation type
  415. dbgln("Found a new exciting relocation type {}", relocation.type());
  416. VERIFY_NOT_REACHED();
  417. }
  418. return RelocationResult::Success;
  419. }
  420. ssize_t DynamicLoader::negative_offset_from_tls_block_end(size_t value_of_symbol, size_t tls_offset, size_t total_tls_size) const
  421. {
  422. auto negative_offset = static_cast<ssize_t>(tls_offset + value_of_symbol - total_tls_size);
  423. // Offset has to be strictly negative. Otherwise we'd collide with the thread's ThreadSpecificData structure.
  424. VERIFY(negative_offset < 0);
  425. return negative_offset;
  426. }
  427. // Defined in <arch>/plt_trampoline.S
  428. extern "C" void _plt_trampoline(void) __attribute__((visibility("hidden")));
  429. void DynamicLoader::setup_plt_trampoline()
  430. {
  431. VERIFY(m_dynamic_object);
  432. VERIFY(m_dynamic_object->has_plt());
  433. VirtualAddress got_address = m_dynamic_object->plt_got_base_address();
  434. auto* got_ptr = (FlatPtr*)got_address.as_ptr();
  435. got_ptr[1] = (FlatPtr)m_dynamic_object.ptr();
  436. got_ptr[2] = (FlatPtr)&_plt_trampoline;
  437. }
  438. // Called from our ASM routine _plt_trampoline.
  439. // Tell the compiler that it might be called from other places:
  440. extern "C" FlatPtr _fixup_plt_entry(DynamicObject* object, u32 relocation_offset);
  441. extern "C" FlatPtr _fixup_plt_entry(DynamicObject* object, u32 relocation_offset)
  442. {
  443. return object->patch_plt_entry(relocation_offset).get();
  444. }
  445. void DynamicLoader::call_object_init_functions()
  446. {
  447. typedef void (*InitFunc)();
  448. if (m_dynamic_object->has_init_section()) {
  449. auto init_function = (InitFunc)(m_dynamic_object->init_section().address().as_ptr());
  450. (init_function)();
  451. }
  452. if (m_dynamic_object->has_init_array_section()) {
  453. auto init_array_section = m_dynamic_object->init_array_section();
  454. InitFunc* init_begin = (InitFunc*)(init_array_section.address().as_ptr());
  455. InitFunc* init_end = init_begin + init_array_section.entry_count();
  456. while (init_begin != init_end) {
  457. // Android sources claim that these can be -1, to be ignored.
  458. // 0 definitely shows up. Apparently 0/-1 are valid? Confusing.
  459. if (!*init_begin || ((FlatPtr)*init_begin == (FlatPtr)-1))
  460. continue;
  461. (*init_begin)();
  462. ++init_begin;
  463. }
  464. }
  465. }
  466. Optional<DynamicObject::SymbolLookupResult> DynamicLoader::lookup_symbol(const ELF::DynamicObject::Symbol& symbol)
  467. {
  468. if (symbol.is_undefined() || symbol.bind() == STB_WEAK)
  469. return DynamicLinker::lookup_global_symbol(symbol.name());
  470. return DynamicObject::SymbolLookupResult { symbol.value(), symbol.address(), symbol.bind(), &symbol.object() };
  471. }
  472. } // end namespace ELF