DynamicLinker.cpp 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465
  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. *
  6. * SPDX-License-Identifier: BSD-2-Clause
  7. */
  8. #include <AK/Debug.h>
  9. #include <AK/HashMap.h>
  10. #include <AK/HashTable.h>
  11. #include <AK/LexicalPath.h>
  12. #include <AK/NonnullRefPtrVector.h>
  13. #include <AK/ScopeGuard.h>
  14. #include <LibC/bits/pthread_integration.h>
  15. #include <LibC/link.h>
  16. #include <LibC/mman.h>
  17. #include <LibC/unistd.h>
  18. #include <LibDl/dlfcn.h>
  19. #include <LibDl/dlfcn_integration.h>
  20. #include <LibELF/AuxiliaryVector.h>
  21. #include <LibELF/DynamicLinker.h>
  22. #include <LibELF/DynamicLoader.h>
  23. #include <LibELF/DynamicObject.h>
  24. #include <LibELF/Hashes.h>
  25. #include <fcntl.h>
  26. #include <sys/types.h>
  27. #include <syscall.h>
  28. namespace ELF {
  29. static HashMap<String, NonnullRefPtr<ELF::DynamicLoader>> s_loaders;
  30. static String s_main_program_name;
  31. static HashMap<String, NonnullRefPtr<ELF::DynamicObject>> s_global_objects;
  32. using EntryPointFunction = int (*)(int, char**, char**);
  33. using LibCExitFunction = void (*)(int);
  34. using DlIteratePhdrCallbackFunction = int (*)(struct dl_phdr_info*, size_t, void*);
  35. using DlIteratePhdrFunction = int (*)(DlIteratePhdrCallbackFunction, void*);
  36. static size_t s_current_tls_offset = 0;
  37. static size_t s_total_tls_size = 0;
  38. static char** s_envp = nullptr;
  39. static LibCExitFunction s_libc_exit = nullptr;
  40. static __pthread_mutex_t s_loader_lock = __PTHREAD_MUTEX_INITIALIZER;
  41. static bool s_allowed_to_check_environment_variables { false };
  42. static bool s_do_breakpoint_trap_before_entry { false };
  43. static Result<void, DlErrorMessage> __dlclose(void* handle);
  44. static Result<void*, DlErrorMessage> __dlopen(const char* filename, int flags);
  45. static Result<void*, DlErrorMessage> __dlsym(void* handle, const char* symbol_name);
  46. Optional<DynamicObject::SymbolLookupResult> DynamicLinker::lookup_global_symbol(const StringView& name)
  47. {
  48. Optional<DynamicObject::SymbolLookupResult> weak_result;
  49. auto symbol = DynamicObject::HashSymbol { name };
  50. for (auto& lib : s_global_objects) {
  51. auto res = lib.value->lookup_symbol(symbol);
  52. if (!res.has_value())
  53. continue;
  54. if (res.value().bind == STB_GLOBAL)
  55. return res;
  56. if (res.value().bind == STB_WEAK && !weak_result.has_value())
  57. weak_result = res;
  58. // We don't want to allow local symbols to be pulled in to other modules
  59. }
  60. return weak_result;
  61. }
  62. static String get_library_name(String path)
  63. {
  64. return LexicalPath(move(path)).basename();
  65. }
  66. static Result<NonnullRefPtr<DynamicLoader>, DlErrorMessage> map_library(const String& filename, int fd, bool is_program)
  67. {
  68. auto result = ELF::DynamicLoader::try_create(fd, filename);
  69. if (result.is_error()) {
  70. return result;
  71. }
  72. auto& loader = result.value();
  73. s_loaders.set(get_library_name(filename), *loader);
  74. if (is_program) {
  75. loader->set_tls_offset(s_current_tls_offset);
  76. s_current_tls_offset += loader->tls_size();
  77. }
  78. return loader;
  79. }
  80. static Result<NonnullRefPtr<DynamicLoader>, DlErrorMessage> map_library(const String& name, bool is_program)
  81. {
  82. if (name.contains("/")) {
  83. int fd = open(name.characters(), O_RDONLY);
  84. if (fd < 0)
  85. return DlErrorMessage { String::formatted("Could not open shared library: {}", name) };
  86. return map_library(name, fd, is_program);
  87. }
  88. // TODO: Do we want to also look for libs in other paths too?
  89. const char* search_paths[] = { "/usr/lib/{}", "/usr/local/lib/{}" };
  90. for (auto& search_path : search_paths) {
  91. auto path = String::formatted(search_path, name);
  92. int fd = open(path.characters(), O_RDONLY);
  93. if (fd < 0)
  94. continue;
  95. return map_library(name, fd, is_program);
  96. }
  97. return DlErrorMessage { String::formatted("Could not find required shared library: {}", name) };
  98. }
  99. static Vector<String> get_dependencies(const String& name)
  100. {
  101. auto lib = s_loaders.get(name).value();
  102. Vector<String> dependencies;
  103. lib->for_each_needed_library([&dependencies, &name](auto needed_name) {
  104. if (name == needed_name)
  105. return IterationDecision::Continue;
  106. dependencies.append(needed_name);
  107. return IterationDecision::Continue;
  108. });
  109. return dependencies;
  110. }
  111. static Result<void, DlErrorMessage> map_dependencies(const String& name)
  112. {
  113. dbgln_if(DYNAMIC_LOAD_DEBUG, "mapping dependencies for: {}", name);
  114. for (const auto& needed_name : get_dependencies(name)) {
  115. dbgln_if(DYNAMIC_LOAD_DEBUG, "needed library: {}", needed_name.characters());
  116. String library_name = get_library_name(needed_name);
  117. if (!s_loaders.contains(library_name) && !s_global_objects.contains(library_name)) {
  118. auto result1 = map_library(needed_name, false);
  119. if (result1.is_error()) {
  120. return result1.error();
  121. }
  122. auto result2 = map_dependencies(library_name);
  123. if (result2.is_error()) {
  124. return result2.error();
  125. }
  126. }
  127. }
  128. dbgln_if(DYNAMIC_LOAD_DEBUG, "mapped dependencies for {}", name);
  129. return {};
  130. }
  131. static void allocate_tls()
  132. {
  133. size_t total_tls_size = 0;
  134. for (const auto& data : s_loaders) {
  135. dbgln_if(DYNAMIC_LOAD_DEBUG, "{}: TLS Size: {}", data.key, data.value->tls_size());
  136. total_tls_size += data.value->tls_size();
  137. }
  138. if (total_tls_size) {
  139. [[maybe_unused]] void* tls_address = ::allocate_tls(total_tls_size);
  140. dbgln_if(DYNAMIC_LOAD_DEBUG, "from userspace, tls_address: {:p}", tls_address);
  141. }
  142. s_total_tls_size = total_tls_size;
  143. }
  144. static int __dl_iterate_phdr(DlIteratePhdrCallbackFunction callback, void* data)
  145. {
  146. __pthread_mutex_lock(&s_loader_lock);
  147. ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
  148. for (auto& it : s_global_objects) {
  149. auto& object = it.value;
  150. auto info = dl_phdr_info {
  151. .dlpi_addr = (ElfW(Addr))object->base_address().as_ptr(),
  152. .dlpi_name = object->filename().characters(),
  153. .dlpi_phdr = object->program_headers(),
  154. .dlpi_phnum = object->program_header_count()
  155. };
  156. auto res = callback(&info, sizeof(info), data);
  157. if (res != 0)
  158. return res;
  159. }
  160. return 0;
  161. }
  162. static void initialize_libc(DynamicObject& libc)
  163. {
  164. // Traditionally, `_start` of the main program initializes libc.
  165. // However, since some libs use malloc() and getenv() in global constructors,
  166. // we have to initialize libc just after it is loaded.
  167. // Also, we can't just mark `__libc_init` with "__attribute__((constructor))"
  168. // because it uses getenv() internally, so `environ` has to be initialized before we call `__libc_init`.
  169. auto res = libc.lookup_symbol("environ"sv);
  170. VERIFY(res.has_value());
  171. *((char***)res.value().address.as_ptr()) = s_envp;
  172. res = libc.lookup_symbol("__environ_is_malloced"sv);
  173. VERIFY(res.has_value());
  174. *((bool*)res.value().address.as_ptr()) = false;
  175. res = libc.lookup_symbol("exit"sv);
  176. VERIFY(res.has_value());
  177. s_libc_exit = (LibCExitFunction)res.value().address.as_ptr();
  178. res = libc.lookup_symbol("__dl_iterate_phdr"sv);
  179. VERIFY(res.has_value());
  180. *((DlIteratePhdrFunction*)res.value().address.as_ptr()) = __dl_iterate_phdr;
  181. res = libc.lookup_symbol("__dlclose"sv);
  182. VERIFY(res.has_value());
  183. *((DlCloseFunction*)res.value().address.as_ptr()) = __dlclose;
  184. res = libc.lookup_symbol("__dlopen"sv);
  185. VERIFY(res.has_value());
  186. *((DlOpenFunction*)res.value().address.as_ptr()) = __dlopen;
  187. res = libc.lookup_symbol("__dlsym"sv);
  188. VERIFY(res.has_value());
  189. *((DlSymFunction*)res.value().address.as_ptr()) = __dlsym;
  190. res = libc.lookup_symbol("__libc_init"sv);
  191. VERIFY(res.has_value());
  192. typedef void libc_init_func();
  193. ((libc_init_func*)res.value().address.as_ptr())();
  194. }
  195. template<typename Callback>
  196. static void for_each_unfinished_dependency_of(const String& name, HashTable<String>& seen_names, Callback callback)
  197. {
  198. if (!s_loaders.contains(name))
  199. return;
  200. if (seen_names.contains(name))
  201. return;
  202. seen_names.set(name);
  203. for (const auto& needed_name : get_dependencies(name))
  204. for_each_unfinished_dependency_of(get_library_name(needed_name), seen_names, callback);
  205. callback(*s_loaders.get(name).value());
  206. }
  207. static NonnullRefPtrVector<DynamicLoader> collect_loaders_for_library(const String& name)
  208. {
  209. HashTable<String> seen_names;
  210. NonnullRefPtrVector<DynamicLoader> loaders;
  211. for_each_unfinished_dependency_of(name, seen_names, [&](auto& loader) {
  212. loaders.append(loader);
  213. });
  214. return loaders;
  215. }
  216. static Result<NonnullRefPtr<DynamicLoader>, DlErrorMessage> load_main_library(const String& name, int flags)
  217. {
  218. auto main_library_loader = *s_loaders.get(name);
  219. auto main_library_object = main_library_loader->map();
  220. s_global_objects.set(name, *main_library_object);
  221. auto loaders = collect_loaders_for_library(name);
  222. for (auto& loader : loaders) {
  223. auto dynamic_object = loader.map();
  224. if (dynamic_object)
  225. s_global_objects.set(dynamic_object->filename(), *dynamic_object);
  226. }
  227. for (auto& loader : loaders) {
  228. bool success = loader.link(flags, s_total_tls_size);
  229. if (!success) {
  230. return DlErrorMessage { String::formatted("Failed to link library {}", loader.filename()) };
  231. }
  232. }
  233. for (auto& loader : loaders) {
  234. auto result = loader.load_stage_3(flags, s_total_tls_size);
  235. VERIFY(!result.is_error());
  236. auto& object = result.value();
  237. if (loader.filename() == "libsystem.so") {
  238. if (syscall(SC_msyscall, object->base_address().as_ptr())) {
  239. VERIFY_NOT_REACHED();
  240. }
  241. }
  242. if (loader.filename() == "libc.so") {
  243. initialize_libc(*object);
  244. }
  245. }
  246. for (auto& loader : loaders) {
  247. loader.load_stage_4();
  248. }
  249. return main_library_loader;
  250. }
  251. static Result<void, DlErrorMessage> __dlclose(void* handle)
  252. {
  253. dbgln_if(DYNAMIC_LOAD_DEBUG, "__dlclose: {}", handle);
  254. __pthread_mutex_lock(&s_loader_lock);
  255. ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
  256. // FIXME: this will not currently destroy the dynamic object
  257. // because we're intentionally holding a strong reference to it
  258. // via s_global_objects until there's proper unload support.
  259. auto object = static_cast<ELF::DynamicObject*>(handle);
  260. object->unref();
  261. return {};
  262. }
  263. static Result<void*, DlErrorMessage> __dlopen(const char* filename, int flags)
  264. {
  265. // FIXME: RTLD_NOW and RTLD_LOCAL are not supported
  266. flags &= ~RTLD_NOW;
  267. flags |= RTLD_LAZY;
  268. flags &= ~RTLD_LOCAL;
  269. flags |= RTLD_GLOBAL;
  270. dbgln_if(DYNAMIC_LOAD_DEBUG, "__dlopen invoked, filename={}, flags={}", filename, flags);
  271. auto library_name = get_library_name(filename ? filename : s_main_program_name);
  272. if (__pthread_mutex_trylock(&s_loader_lock) != 0)
  273. return DlErrorMessage { "Nested calls to dlopen() are not permitted." };
  274. ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
  275. auto existing_elf_object = s_global_objects.get(library_name);
  276. if (existing_elf_object.has_value()) {
  277. // It's up to the caller to release the ref with dlclose().
  278. return &existing_elf_object->leak_ref();
  279. }
  280. VERIFY(!library_name.is_empty());
  281. auto result1 = map_library(filename, false);
  282. if (result1.is_error()) {
  283. return result1.error();
  284. }
  285. auto result2 = map_dependencies(library_name);
  286. if (result2.is_error()) {
  287. return result2.error();
  288. }
  289. auto result = load_main_library(library_name, flags);
  290. if (result.is_error())
  291. return result.error();
  292. auto object = s_global_objects.get(library_name);
  293. if (!object.has_value())
  294. return DlErrorMessage { "Could not load ELF object." };
  295. // It's up to the caller to release the ref with dlclose().
  296. return &object->leak_ref();
  297. }
  298. static Result<void*, DlErrorMessage> __dlsym(void* handle, const char* symbol_name)
  299. {
  300. dbgln_if(DYNAMIC_LOAD_DEBUG, "__dlsym: {}, {}", handle, symbol_name);
  301. __pthread_mutex_lock(&s_loader_lock);
  302. ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
  303. auto object = static_cast<DynamicObject*>(handle);
  304. if (!handle) {
  305. auto library_name = get_library_name(s_main_program_name);
  306. auto global_object = s_global_objects.get(library_name);
  307. object = *global_object;
  308. }
  309. auto symbol = object->lookup_symbol(symbol_name);
  310. if (!symbol.has_value()) {
  311. return DlErrorMessage { String::formatted("Symbol {} not found", symbol_name) };
  312. }
  313. return symbol.value().address.as_ptr();
  314. }
  315. static void read_environment_variables()
  316. {
  317. for (char** env = s_envp; *env; ++env) {
  318. if (StringView { *env } == "_LOADER_BREAKPOINT=1") {
  319. s_do_breakpoint_trap_before_entry = true;
  320. }
  321. }
  322. }
  323. void ELF::DynamicLinker::linker_main(String&& main_program_name, int main_program_fd, bool is_secure, int argc, char** argv, char** envp)
  324. {
  325. s_envp = envp;
  326. s_allowed_to_check_environment_variables = !is_secure;
  327. if (s_allowed_to_check_environment_variables)
  328. read_environment_variables();
  329. s_main_program_name = main_program_name;
  330. auto library_name = get_library_name(main_program_name);
  331. // NOTE: We always map the main library first, since it may require
  332. // placement at a specific address.
  333. auto result1 = map_library(main_program_name, main_program_fd, true);
  334. if (result1.is_error()) {
  335. warnln("{}", result1.error().text);
  336. fflush(stderr);
  337. _exit(1);
  338. }
  339. result1.release_value();
  340. auto result2 = map_dependencies(library_name);
  341. if (result2.is_error()) {
  342. warnln("{}", result2.error().text);
  343. fflush(stderr);
  344. _exit(1);
  345. }
  346. dbgln_if(DYNAMIC_LOAD_DEBUG, "loaded all dependencies");
  347. for ([[maybe_unused]] auto& lib : s_loaders) {
  348. dbgln_if(DYNAMIC_LOAD_DEBUG, "{} - tls size: {}, tls offset: {}", lib.key, lib.value->tls_size(), lib.value->tls_offset());
  349. }
  350. allocate_tls();
  351. auto entry_point_function = [&main_program_name] {
  352. auto library_name = get_library_name(main_program_name);
  353. auto result = load_main_library(library_name, RTLD_GLOBAL | RTLD_LAZY);
  354. if (result.is_error()) {
  355. warnln("{}", result.error().text);
  356. _exit(1);
  357. }
  358. auto& main_executable_loader = result.value();
  359. auto entry_point = main_executable_loader->image().entry();
  360. if (main_executable_loader->is_dynamic())
  361. entry_point = entry_point.offset(main_executable_loader->base_address().get());
  362. return (EntryPointFunction)(entry_point.as_ptr());
  363. }();
  364. s_loaders.clear();
  365. int rc = syscall(SC_msyscall, nullptr);
  366. if (rc < 0) {
  367. VERIFY_NOT_REACHED();
  368. }
  369. dbgln_if(DYNAMIC_LOAD_DEBUG, "Jumping to entry point: {:p}", entry_point_function);
  370. if (s_do_breakpoint_trap_before_entry) {
  371. asm("int3");
  372. }
  373. rc = entry_point_function(argc, argv, envp);
  374. dbgln_if(DYNAMIC_LOAD_DEBUG, "rc: {}", rc);
  375. if (s_libc_exit != nullptr) {
  376. s_libc_exit(rc);
  377. } else {
  378. _exit(rc);
  379. }
  380. VERIFY_NOT_REACHED();
  381. }
  382. }