DynamicLinker.cpp 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589
  1. /*
  2. * Copyright (c) 2020, Itamar S. <itamar8910@gmail.com>
  3. * Copyright (c) 2021, Andreas Kling <kling@serenityos.org>
  4. * Copyright (c) 2021, the SerenityOS developers.
  5. * Copyright (c) 2022, Jesse Buhagiar <jooster669@gmail.com>
  6. *
  7. * SPDX-License-Identifier: BSD-2-Clause
  8. */
  9. #include <AK/ByteBuffer.h>
  10. #include <AK/Debug.h>
  11. #include <AK/HashMap.h>
  12. #include <AK/HashTable.h>
  13. #include <AK/LexicalPath.h>
  14. #include <AK/NonnullRefPtrVector.h>
  15. #include <AK/ScopeGuard.h>
  16. #include <AK/Vector.h>
  17. #include <LibC/bits/pthread_integration.h>
  18. #include <LibC/link.h>
  19. #include <LibC/sys/mman.h>
  20. #include <LibC/unistd.h>
  21. #include <LibDl/dlfcn.h>
  22. #include <LibDl/dlfcn_integration.h>
  23. #include <LibELF/AuxiliaryVector.h>
  24. #include <LibELF/DynamicLinker.h>
  25. #include <LibELF/DynamicLoader.h>
  26. #include <LibELF/DynamicObject.h>
  27. #include <LibELF/Hashes.h>
  28. #include <fcntl.h>
  29. #include <string.h>
  30. #include <sys/types.h>
  31. #include <syscall.h>
  32. namespace ELF {
  33. static HashMap<String, NonnullRefPtr<ELF::DynamicLoader>> s_loaders;
  34. static String s_main_program_name;
  35. static OrderedHashMap<String, NonnullRefPtr<ELF::DynamicObject>> s_global_objects;
  36. using EntryPointFunction = int (*)(int, char**, char**);
  37. using LibCExitFunction = void (*)(int);
  38. using DlIteratePhdrCallbackFunction = int (*)(struct dl_phdr_info*, size_t, void*);
  39. using DlIteratePhdrFunction = int (*)(DlIteratePhdrCallbackFunction, void*);
  40. extern "C" [[noreturn]] void _invoke_entry(int argc, char** argv, char** envp, EntryPointFunction entry);
  41. static size_t s_current_tls_offset = 0;
  42. static size_t s_total_tls_size = 0;
  43. static size_t s_allocated_tls_block_size = 0;
  44. static char** s_envp = nullptr;
  45. static LibCExitFunction s_libc_exit = nullptr;
  46. static __pthread_mutex_t s_loader_lock = __PTHREAD_MUTEX_INITIALIZER;
  47. static bool s_allowed_to_check_environment_variables { false };
  48. static bool s_do_breakpoint_trap_before_entry { false };
  49. static StringView s_ld_library_path;
  50. static Result<void, DlErrorMessage> __dlclose(void* handle);
  51. static Result<void*, DlErrorMessage> __dlopen(const char* filename, int flags);
  52. static Result<void*, DlErrorMessage> __dlsym(void* handle, const char* symbol_name);
  53. static Result<void, DlErrorMessage> __dladdr(void* addr, Dl_info* info);
  54. Optional<DynamicObject::SymbolLookupResult> DynamicLinker::lookup_global_symbol(StringView name)
  55. {
  56. Optional<DynamicObject::SymbolLookupResult> weak_result;
  57. auto symbol = DynamicObject::HashSymbol { name };
  58. for (auto& lib : s_global_objects) {
  59. auto res = lib.value->lookup_symbol(symbol);
  60. if (!res.has_value())
  61. continue;
  62. if (res.value().bind == STB_GLOBAL)
  63. return res;
  64. if (res.value().bind == STB_WEAK && !weak_result.has_value())
  65. weak_result = res;
  66. // We don't want to allow local symbols to be pulled in to other modules
  67. }
  68. return weak_result;
  69. }
  70. static String get_library_name(String path)
  71. {
  72. return LexicalPath::basename(move(path));
  73. }
  74. static Result<NonnullRefPtr<DynamicLoader>, DlErrorMessage> map_library(const String& filename, int fd)
  75. {
  76. auto result = ELF::DynamicLoader::try_create(fd, filename);
  77. if (result.is_error()) {
  78. return result;
  79. }
  80. auto& loader = result.value();
  81. s_loaders.set(get_library_name(filename), *loader);
  82. s_current_tls_offset -= loader->tls_size_of_current_object();
  83. loader->set_tls_offset(s_current_tls_offset);
  84. return loader;
  85. }
  86. static Result<NonnullRefPtr<DynamicLoader>, DlErrorMessage> map_library(const String& name)
  87. {
  88. if (name.contains("/"sv)) {
  89. int fd = open(name.characters(), O_RDONLY);
  90. if (fd < 0)
  91. return DlErrorMessage { String::formatted("Could not open shared library: {}", name) };
  92. return map_library(name, fd);
  93. }
  94. // Scan the LD_LIBRARY_PATH environment variable if applicable
  95. if (s_ld_library_path != nullptr) {
  96. for (const auto& search_path : s_ld_library_path.split_view(':')) {
  97. LexicalPath library_path(search_path);
  98. int fd = open(library_path.append(name).string().characters(), O_RDONLY);
  99. if (fd < 0)
  100. continue;
  101. return map_library(name, fd);
  102. }
  103. }
  104. // Now check the default paths.
  105. // TODO: Do we want to also look for libs in other paths too?
  106. const char* search_paths[] = { "/usr/lib/{}", "/usr/local/lib/{}" };
  107. for (auto& search_path : search_paths) {
  108. auto path = String::formatted(search_path, name);
  109. int fd = open(path.characters(), O_RDONLY);
  110. if (fd < 0)
  111. continue;
  112. return map_library(name, fd);
  113. }
  114. return DlErrorMessage { String::formatted("Could not find required shared library: {}", name) };
  115. }
  116. static Vector<String> get_dependencies(const String& name)
  117. {
  118. auto lib = s_loaders.get(name).value();
  119. Vector<String> dependencies;
  120. lib->for_each_needed_library([&dependencies, &name](auto needed_name) {
  121. if (name == needed_name)
  122. return;
  123. dependencies.append(needed_name);
  124. });
  125. return dependencies;
  126. }
  127. static Result<void, DlErrorMessage> map_dependencies(const String& name)
  128. {
  129. dbgln_if(DYNAMIC_LOAD_DEBUG, "mapping dependencies for: {}", name);
  130. for (const auto& needed_name : get_dependencies(name)) {
  131. dbgln_if(DYNAMIC_LOAD_DEBUG, "needed library: {}", needed_name.characters());
  132. String library_name = get_library_name(needed_name);
  133. if (!s_loaders.contains(library_name) && !s_global_objects.contains(library_name)) {
  134. auto result1 = map_library(needed_name);
  135. if (result1.is_error()) {
  136. return result1.error();
  137. }
  138. auto result2 = map_dependencies(library_name);
  139. if (result2.is_error()) {
  140. return result2.error();
  141. }
  142. }
  143. }
  144. dbgln_if(DYNAMIC_LOAD_DEBUG, "mapped dependencies for {}", name);
  145. return {};
  146. }
  147. static void allocate_tls()
  148. {
  149. s_total_tls_size = 0;
  150. for (const auto& data : s_loaders) {
  151. dbgln_if(DYNAMIC_LOAD_DEBUG, "{}: TLS Size: {}", data.key, data.value->tls_size_of_current_object());
  152. s_total_tls_size += data.value->tls_size_of_current_object();
  153. }
  154. if (!s_total_tls_size)
  155. return;
  156. auto page_aligned_size = align_up_to(s_total_tls_size, PAGE_SIZE);
  157. auto initial_tls_data_result = ByteBuffer::create_zeroed(page_aligned_size);
  158. if (!initial_tls_data_result.has_value()) {
  159. dbgln("Failed to allocate initial TLS data");
  160. VERIFY_NOT_REACHED();
  161. }
  162. auto& initial_tls_data = initial_tls_data_result.value();
  163. // Initialize TLS data
  164. for (const auto& entry : s_loaders) {
  165. entry.value->copy_initial_tls_data_into(initial_tls_data);
  166. }
  167. void* master_tls = ::allocate_tls((char*)initial_tls_data.data(), initial_tls_data.size());
  168. VERIFY(master_tls != (void*)-1);
  169. dbgln_if(DYNAMIC_LOAD_DEBUG, "from userspace, master_tls: {:p}", master_tls);
  170. s_allocated_tls_block_size = initial_tls_data.size();
  171. }
  172. static int __dl_iterate_phdr(DlIteratePhdrCallbackFunction callback, void* data)
  173. {
  174. __pthread_mutex_lock(&s_loader_lock);
  175. ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
  176. for (auto& it : s_global_objects) {
  177. auto& object = it.value;
  178. auto info = dl_phdr_info {
  179. .dlpi_addr = (ElfW(Addr))object->base_address().as_ptr(),
  180. .dlpi_name = object->filename().characters(),
  181. .dlpi_phdr = object->program_headers(),
  182. .dlpi_phnum = object->program_header_count()
  183. };
  184. auto res = callback(&info, sizeof(info), data);
  185. if (res != 0)
  186. return res;
  187. }
  188. return 0;
  189. }
  190. static void initialize_libc(DynamicObject& libc)
  191. {
  192. // Traditionally, `_start` of the main program initializes libc.
  193. // However, since some libs use malloc() and getenv() in global constructors,
  194. // we have to initialize libc just after it is loaded.
  195. // Also, we can't just mark `__libc_init` with "__attribute__((constructor))"
  196. // because it uses getenv() internally, so `environ` has to be initialized before we call `__libc_init`.
  197. auto res = libc.lookup_symbol("environ"sv);
  198. VERIFY(res.has_value());
  199. *((char***)res.value().address.as_ptr()) = s_envp;
  200. res = libc.lookup_symbol("__environ_is_malloced"sv);
  201. VERIFY(res.has_value());
  202. *((bool*)res.value().address.as_ptr()) = false;
  203. res = libc.lookup_symbol("exit"sv);
  204. VERIFY(res.has_value());
  205. s_libc_exit = (LibCExitFunction)res.value().address.as_ptr();
  206. res = libc.lookup_symbol("__dl_iterate_phdr"sv);
  207. VERIFY(res.has_value());
  208. *((DlIteratePhdrFunction*)res.value().address.as_ptr()) = __dl_iterate_phdr;
  209. res = libc.lookup_symbol("__dlclose"sv);
  210. VERIFY(res.has_value());
  211. *((DlCloseFunction*)res.value().address.as_ptr()) = __dlclose;
  212. res = libc.lookup_symbol("__dlopen"sv);
  213. VERIFY(res.has_value());
  214. *((DlOpenFunction*)res.value().address.as_ptr()) = __dlopen;
  215. res = libc.lookup_symbol("__dlsym"sv);
  216. VERIFY(res.has_value());
  217. *((DlSymFunction*)res.value().address.as_ptr()) = __dlsym;
  218. res = libc.lookup_symbol("__dladdr"sv);
  219. VERIFY(res.has_value());
  220. *((DlAddrFunction*)res.value().address.as_ptr()) = __dladdr;
  221. res = libc.lookup_symbol("__libc_init"sv);
  222. VERIFY(res.has_value());
  223. typedef void libc_init_func();
  224. ((libc_init_func*)res.value().address.as_ptr())();
  225. }
  226. template<typename Callback>
  227. static void for_each_unfinished_dependency_of(const String& name, HashTable<String>& seen_names, bool first, bool skip_global_objects, Callback callback)
  228. {
  229. if (!s_loaders.contains(name))
  230. return;
  231. if (!first && skip_global_objects && s_global_objects.contains(name))
  232. return;
  233. if (seen_names.contains(name))
  234. return;
  235. seen_names.set(name);
  236. for (const auto& needed_name : get_dependencies(name))
  237. for_each_unfinished_dependency_of(get_library_name(needed_name), seen_names, false, skip_global_objects, callback);
  238. callback(*s_loaders.get(name).value());
  239. }
  240. static NonnullRefPtrVector<DynamicLoader> collect_loaders_for_library(const String& name, bool skip_global_objects)
  241. {
  242. HashTable<String> seen_names;
  243. NonnullRefPtrVector<DynamicLoader> loaders;
  244. for_each_unfinished_dependency_of(name, seen_names, true, skip_global_objects, [&](auto& loader) {
  245. loaders.append(loader);
  246. });
  247. return loaders;
  248. }
  249. static Result<NonnullRefPtr<DynamicLoader>, DlErrorMessage> load_main_library(const String& name, int flags, bool skip_global_objects)
  250. {
  251. auto main_library_loader = *s_loaders.get(name);
  252. auto main_library_object = main_library_loader->map();
  253. s_global_objects.set(name, *main_library_object);
  254. auto loaders = collect_loaders_for_library(name, skip_global_objects);
  255. for (auto& loader : loaders) {
  256. auto dynamic_object = loader.map();
  257. if (dynamic_object)
  258. s_global_objects.set(dynamic_object->filename(), *dynamic_object);
  259. }
  260. for (auto& loader : loaders) {
  261. bool success = loader.link(flags);
  262. if (!success) {
  263. return DlErrorMessage { String::formatted("Failed to link library {}", loader.filename()) };
  264. }
  265. }
  266. for (auto& loader : loaders) {
  267. auto result = loader.load_stage_3(flags);
  268. VERIFY(!result.is_error());
  269. auto& object = result.value();
  270. if (loader.filename() == "libsystem.so"sv) {
  271. VERIFY(!loader.text_segments().is_empty());
  272. for (const auto& segment : loader.text_segments()) {
  273. if (syscall(SC_msyscall, segment.address().get())) {
  274. VERIFY_NOT_REACHED();
  275. }
  276. }
  277. }
  278. if (loader.filename() == "libc.so"sv) {
  279. initialize_libc(*object);
  280. }
  281. }
  282. for (auto& loader : loaders) {
  283. loader.load_stage_4();
  284. }
  285. return NonnullRefPtr<DynamicLoader>(*main_library_loader);
  286. }
  287. static Result<void, DlErrorMessage> __dlclose(void* handle)
  288. {
  289. dbgln_if(DYNAMIC_LOAD_DEBUG, "__dlclose: {}", handle);
  290. __pthread_mutex_lock(&s_loader_lock);
  291. ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
  292. // FIXME: this will not currently destroy the dynamic object
  293. // because we're intentionally holding a strong reference to it
  294. // via s_global_objects until there's proper unload support.
  295. auto object = static_cast<ELF::DynamicObject*>(handle);
  296. object->unref();
  297. return {};
  298. }
  299. static Optional<DlErrorMessage> verify_tls_for_dlopen(const DynamicLoader& loader)
  300. {
  301. if (loader.tls_size_of_current_object() == 0)
  302. return {};
  303. if (s_total_tls_size + loader.tls_size_of_current_object() > s_allocated_tls_block_size)
  304. return DlErrorMessage("TLS size too large");
  305. bool tls_data_is_all_zero = true;
  306. loader.image().for_each_program_header([&loader, &tls_data_is_all_zero](ELF::Image::ProgramHeader program_header) {
  307. if (program_header.type() != PT_TLS)
  308. return IterationDecision::Continue;
  309. auto* tls_data = (const u8*)loader.image().base_address() + program_header.offset();
  310. for (size_t i = 0; i < program_header.size_in_image(); ++i) {
  311. if (tls_data[i] != 0) {
  312. tls_data_is_all_zero = false;
  313. break;
  314. }
  315. }
  316. return IterationDecision::Break;
  317. });
  318. if (tls_data_is_all_zero)
  319. return {};
  320. return DlErrorMessage("Using dlopen() with libraries that have non-zeroed TLS is currently not supported");
  321. }
  322. static Result<void*, DlErrorMessage> __dlopen(const char* filename, int flags)
  323. {
  324. // FIXME: RTLD_NOW and RTLD_LOCAL are not supported
  325. flags &= ~RTLD_NOW;
  326. flags |= RTLD_LAZY;
  327. flags &= ~RTLD_LOCAL;
  328. flags |= RTLD_GLOBAL;
  329. dbgln_if(DYNAMIC_LOAD_DEBUG, "__dlopen invoked, filename={}, flags={}", filename, flags);
  330. auto library_name = get_library_name(filename ? filename : s_main_program_name);
  331. if (__pthread_mutex_trylock(&s_loader_lock) != 0)
  332. return DlErrorMessage { "Nested calls to dlopen() are not permitted." };
  333. ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
  334. auto existing_elf_object = s_global_objects.get(library_name);
  335. if (existing_elf_object.has_value()) {
  336. // It's up to the caller to release the ref with dlclose().
  337. existing_elf_object.value()->ref();
  338. return *existing_elf_object;
  339. }
  340. VERIFY(!library_name.is_empty());
  341. auto result1 = map_library(filename);
  342. if (result1.is_error()) {
  343. return result1.error();
  344. }
  345. if (auto error = verify_tls_for_dlopen(result1.value()); error.has_value())
  346. return error.value();
  347. auto result2 = map_dependencies(library_name);
  348. if (result2.is_error()) {
  349. return result2.error();
  350. }
  351. auto result = load_main_library(library_name, flags, true);
  352. if (result.is_error())
  353. return result.error();
  354. s_total_tls_size += result1.value()->tls_size_of_current_object();
  355. auto object = s_global_objects.get(library_name);
  356. if (!object.has_value())
  357. return DlErrorMessage { "Could not load ELF object." };
  358. // It's up to the caller to release the ref with dlclose().
  359. object.value()->ref();
  360. return *object;
  361. }
  362. static Result<void*, DlErrorMessage> __dlsym(void* handle, const char* symbol_name)
  363. {
  364. dbgln_if(DYNAMIC_LOAD_DEBUG, "__dlsym: {}, {}", handle, symbol_name);
  365. __pthread_mutex_lock(&s_loader_lock);
  366. ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
  367. auto object = static_cast<DynamicObject*>(handle);
  368. if (!handle) {
  369. auto library_name = get_library_name(s_main_program_name);
  370. auto global_object = s_global_objects.get(library_name);
  371. object = *global_object;
  372. }
  373. auto symbol = object->lookup_symbol(symbol_name);
  374. if (!symbol.has_value()) {
  375. return DlErrorMessage { String::formatted("Symbol {} not found", symbol_name) };
  376. }
  377. return symbol.value().address.as_ptr();
  378. }
  379. static Result<void, DlErrorMessage> __dladdr(void* addr, Dl_info* info)
  380. {
  381. VirtualAddress user_addr { addr };
  382. __pthread_mutex_lock(&s_loader_lock);
  383. ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
  384. RefPtr<DynamicObject> best_matching_library;
  385. VirtualAddress best_library_offset;
  386. for (auto& lib : s_global_objects) {
  387. if (user_addr < lib.value->base_address())
  388. continue;
  389. auto offset = user_addr - lib.value->base_address();
  390. if (!best_matching_library || offset < best_library_offset) {
  391. best_matching_library = lib.value;
  392. best_library_offset = offset;
  393. }
  394. }
  395. if (!best_matching_library) {
  396. return DlErrorMessage { "No library found which contains the specified address" };
  397. }
  398. Optional<DynamicObject::Symbol> best_matching_symbol;
  399. best_matching_library->for_each_symbol([&](auto const& symbol) {
  400. if (user_addr < symbol.address() || user_addr > symbol.address().offset(symbol.size()))
  401. return;
  402. best_matching_symbol = symbol;
  403. });
  404. info->dli_fbase = best_matching_library->base_address().as_ptr();
  405. // This works because we don't support unloading objects.
  406. info->dli_fname = best_matching_library->filename().characters();
  407. if (best_matching_symbol.has_value()) {
  408. info->dli_saddr = best_matching_symbol.value().address().as_ptr();
  409. info->dli_sname = best_matching_symbol.value().raw_name();
  410. } else {
  411. info->dli_saddr = nullptr;
  412. info->dli_sname = nullptr;
  413. }
  414. return {};
  415. }
  416. static void read_environment_variables()
  417. {
  418. for (char** env = s_envp; *env; ++env) {
  419. StringView env_string { *env };
  420. if (env_string == "_LOADER_BREAKPOINT=1"sv) {
  421. s_do_breakpoint_trap_before_entry = true;
  422. }
  423. constexpr auto library_path_string = "LD_LIBRARY_PATH="sv;
  424. if (env_string.starts_with(library_path_string)) {
  425. s_ld_library_path = env_string.substring_view(library_path_string.length());
  426. }
  427. }
  428. }
  429. void ELF::DynamicLinker::linker_main(String&& main_program_name, int main_program_fd, bool is_secure, int argc, char** argv, char** envp)
  430. {
  431. s_envp = envp;
  432. s_allowed_to_check_environment_variables = !is_secure;
  433. if (s_allowed_to_check_environment_variables)
  434. read_environment_variables();
  435. s_main_program_name = main_program_name;
  436. auto library_name = get_library_name(main_program_name);
  437. // NOTE: We always map the main library first, since it may require
  438. // placement at a specific address.
  439. auto result1 = map_library(main_program_name, main_program_fd);
  440. if (result1.is_error()) {
  441. warnln("{}", result1.error().text);
  442. fflush(stderr);
  443. _exit(1);
  444. }
  445. (void)result1.release_value();
  446. auto result2 = map_dependencies(library_name);
  447. if (result2.is_error()) {
  448. warnln("{}", result2.error().text);
  449. fflush(stderr);
  450. _exit(1);
  451. }
  452. dbgln_if(DYNAMIC_LOAD_DEBUG, "loaded all dependencies");
  453. for ([[maybe_unused]] auto& lib : s_loaders) {
  454. dbgln_if(DYNAMIC_LOAD_DEBUG, "{} - tls size: {}, tls offset: {}", lib.key, lib.value->tls_size_of_current_object(), lib.value->tls_offset());
  455. }
  456. allocate_tls();
  457. auto entry_point_function = [&main_program_name] {
  458. auto library_name = get_library_name(main_program_name);
  459. auto result = load_main_library(library_name, RTLD_GLOBAL | RTLD_LAZY, false);
  460. if (result.is_error()) {
  461. warnln("{}", result.error().text);
  462. _exit(1);
  463. }
  464. auto& main_executable_loader = result.value();
  465. auto entry_point = main_executable_loader->image().entry();
  466. if (main_executable_loader->is_dynamic())
  467. entry_point = entry_point.offset(main_executable_loader->base_address().get());
  468. return (EntryPointFunction)(entry_point.as_ptr());
  469. }();
  470. s_loaders.clear();
  471. int rc = syscall(SC_msyscall, nullptr);
  472. if (rc < 0) {
  473. VERIFY_NOT_REACHED();
  474. }
  475. dbgln_if(DYNAMIC_LOAD_DEBUG, "Jumping to entry point: {:p}", entry_point_function);
  476. if (s_do_breakpoint_trap_before_entry) {
  477. asm("int3");
  478. }
  479. _invoke_entry(argc, argv, envp, entry_point_function);
  480. VERIFY_NOT_REACHED();
  481. }
  482. }