DynamicLinker.cpp 9.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317
  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/link.h>
  15. #include <LibC/mman.h>
  16. #include <LibC/unistd.h>
  17. #include <LibELF/AuxiliaryVector.h>
  18. #include <LibELF/DynamicLinker.h>
  19. #include <LibELF/DynamicLoader.h>
  20. #include <LibELF/DynamicObject.h>
  21. #include <LibELF/Hashes.h>
  22. #include <dlfcn.h>
  23. #include <fcntl.h>
  24. #include <sys/types.h>
  25. #include <syscall.h>
  26. namespace ELF {
  27. namespace {
  28. HashMap<String, NonnullRefPtr<ELF::DynamicLoader>> g_loaders;
  29. Vector<NonnullRefPtr<ELF::DynamicObject>> g_global_objects;
  30. using EntryPointFunction = int (*)(int, char**, char**);
  31. using LibCExitFunction = void (*)(int);
  32. using DlIteratePhdrCallbackFunction = int (*)(struct dl_phdr_info*, size_t, void*);
  33. using DlIteratePhdrFunction = int (*)(DlIteratePhdrCallbackFunction, void*);
  34. size_t g_current_tls_offset = 0;
  35. size_t g_total_tls_size = 0;
  36. char** g_envp = nullptr;
  37. LibCExitFunction g_libc_exit = nullptr;
  38. bool g_allowed_to_check_environment_variables { false };
  39. bool g_do_breakpoint_trap_before_entry { false };
  40. }
  41. Optional<DynamicObject::SymbolLookupResult> DynamicLinker::lookup_global_symbol(const StringView& name)
  42. {
  43. Optional<DynamicObject::SymbolLookupResult> weak_result;
  44. auto symbol = DynamicObject::HashSymbol { name };
  45. for (auto& lib : g_global_objects) {
  46. auto res = lib->lookup_symbol(symbol);
  47. if (!res.has_value())
  48. continue;
  49. if (res.value().bind == STB_GLOBAL)
  50. return res;
  51. if (res.value().bind == STB_WEAK && !weak_result.has_value())
  52. weak_result = res;
  53. // We don't want to allow local symbols to be pulled in to other modules
  54. }
  55. return weak_result;
  56. }
  57. static void map_library(const String& name, int fd)
  58. {
  59. auto loader = ELF::DynamicLoader::try_create(fd, name);
  60. if (!loader) {
  61. dbgln("Failed to create ELF::DynamicLoader for fd={}, name={}", fd, name);
  62. VERIFY_NOT_REACHED();
  63. }
  64. loader->set_tls_offset(g_current_tls_offset);
  65. g_loaders.set(name, *loader);
  66. g_current_tls_offset += loader->tls_size();
  67. }
  68. static void map_library(const String& name)
  69. {
  70. // TODO: Do we want to also look for libs in other paths too?
  71. const char* search_paths[] = { "/usr/lib/{}", "/usr/local/lib/{}" };
  72. for (auto& search_path : search_paths) {
  73. auto path = String::formatted(search_path, name);
  74. int fd = open(path.characters(), O_RDONLY);
  75. if (fd < 0)
  76. continue;
  77. map_library(name, fd);
  78. return;
  79. }
  80. fprintf(stderr, "Could not find required shared library: %s\n", name.characters());
  81. VERIFY_NOT_REACHED();
  82. }
  83. static String get_library_name(String path)
  84. {
  85. return LexicalPath(move(path)).basename();
  86. }
  87. static Vector<String> get_dependencies(const String& name)
  88. {
  89. auto lib = g_loaders.get(name).value();
  90. Vector<String> dependencies;
  91. lib->for_each_needed_library([&dependencies, &name](auto needed_name) {
  92. if (name == needed_name)
  93. return IterationDecision::Continue;
  94. dependencies.append(needed_name);
  95. return IterationDecision::Continue;
  96. });
  97. return dependencies;
  98. }
  99. static void map_dependencies(const String& name)
  100. {
  101. dbgln_if(DYNAMIC_LOAD_DEBUG, "mapping dependencies for: {}", name);
  102. for (const auto& needed_name : get_dependencies(name)) {
  103. dbgln_if(DYNAMIC_LOAD_DEBUG, "needed library: {}", needed_name.characters());
  104. String library_name = get_library_name(needed_name);
  105. if (!g_loaders.contains(library_name)) {
  106. map_library(library_name);
  107. map_dependencies(library_name);
  108. }
  109. }
  110. dbgln_if(DYNAMIC_LOAD_DEBUG, "mapped dependencies for {}", name);
  111. }
  112. static void allocate_tls()
  113. {
  114. size_t total_tls_size = 0;
  115. for (const auto& data : g_loaders) {
  116. dbgln_if(DYNAMIC_LOAD_DEBUG, "{}: TLS Size: {}", data.key, data.value->tls_size());
  117. total_tls_size += data.value->tls_size();
  118. }
  119. if (total_tls_size) {
  120. [[maybe_unused]] void* tls_address = ::allocate_tls(total_tls_size);
  121. dbgln_if(DYNAMIC_LOAD_DEBUG, "from userspace, tls_address: {:p}", tls_address);
  122. }
  123. g_total_tls_size = total_tls_size;
  124. }
  125. static int __dl_iterate_phdr(DlIteratePhdrCallbackFunction callback, void* data)
  126. {
  127. for (auto& object : g_global_objects) {
  128. auto info = dl_phdr_info {
  129. .dlpi_addr = (ElfW(Addr))object->base_address().as_ptr(),
  130. .dlpi_name = object->filename().characters(),
  131. .dlpi_phdr = object->program_headers(),
  132. .dlpi_phnum = object->program_header_count()
  133. };
  134. auto res = callback(&info, sizeof(info), data);
  135. if (res != 0)
  136. return res;
  137. }
  138. return 0;
  139. }
  140. static void initialize_libc(DynamicObject& libc)
  141. {
  142. // Traditionally, `_start` of the main program initializes libc.
  143. // However, since some libs use malloc() and getenv() in global constructors,
  144. // we have to initialize libc just after it is loaded.
  145. // Also, we can't just mark `__libc_init` with "__attribute__((constructor))"
  146. // because it uses getenv() internally, so `environ` has to be initialized before we call `__libc_init`.
  147. auto res = libc.lookup_symbol("environ"sv);
  148. VERIFY(res.has_value());
  149. *((char***)res.value().address.as_ptr()) = g_envp;
  150. res = libc.lookup_symbol("__environ_is_malloced"sv);
  151. VERIFY(res.has_value());
  152. *((bool*)res.value().address.as_ptr()) = false;
  153. res = libc.lookup_symbol("exit"sv);
  154. VERIFY(res.has_value());
  155. g_libc_exit = (LibCExitFunction)res.value().address.as_ptr();
  156. res = libc.lookup_symbol("__dl_iterate_phdr"sv);
  157. VERIFY(res.has_value());
  158. *((DlIteratePhdrFunction*)res.value().address.as_ptr()) = __dl_iterate_phdr;
  159. res = libc.lookup_symbol("__libc_init"sv);
  160. VERIFY(res.has_value());
  161. typedef void libc_init_func();
  162. ((libc_init_func*)res.value().address.as_ptr())();
  163. }
  164. template<typename Callback>
  165. static void for_each_dependency_of(const String& name, HashTable<String>& seen_names, Callback callback)
  166. {
  167. if (seen_names.contains(name))
  168. return;
  169. seen_names.set(name);
  170. for (const auto& needed_name : get_dependencies(name))
  171. for_each_dependency_of(get_library_name(needed_name), seen_names, callback);
  172. callback(*g_loaders.get(name).value());
  173. }
  174. static NonnullRefPtrVector<DynamicLoader> collect_loaders_for_executable(const String& name)
  175. {
  176. HashTable<String> seen_names;
  177. NonnullRefPtrVector<DynamicLoader> loaders;
  178. for_each_dependency_of(name, seen_names, [&](auto& loader) {
  179. loaders.append(loader);
  180. });
  181. return loaders;
  182. }
  183. static NonnullRefPtr<DynamicLoader> load_main_executable(const String& name)
  184. {
  185. // NOTE: We always map the main executable first, since it may require
  186. // placement at a specific address.
  187. auto& main_executable_loader = *g_loaders.get(name).value();
  188. auto main_executable_object = main_executable_loader.map();
  189. g_global_objects.append(*main_executable_object);
  190. auto loaders = collect_loaders_for_executable(name);
  191. for (auto& loader : loaders) {
  192. auto dynamic_object = loader.map();
  193. if (dynamic_object)
  194. g_global_objects.append(*dynamic_object);
  195. }
  196. for (auto& loader : loaders) {
  197. bool success = loader.link(RTLD_GLOBAL | RTLD_LAZY, g_total_tls_size);
  198. VERIFY(success);
  199. }
  200. for (auto& loader : loaders) {
  201. auto object = loader.load_stage_3(RTLD_GLOBAL | RTLD_LAZY, g_total_tls_size);
  202. VERIFY(object);
  203. if (loader.filename() == "libsystem.so") {
  204. if (syscall(SC_msyscall, object->base_address().as_ptr())) {
  205. VERIFY_NOT_REACHED();
  206. }
  207. }
  208. if (loader.filename() == "libc.so") {
  209. initialize_libc(*object);
  210. }
  211. }
  212. for (auto& loader : loaders) {
  213. loader.load_stage_4();
  214. }
  215. return main_executable_loader;
  216. }
  217. static void read_environment_variables()
  218. {
  219. for (char** env = g_envp; *env; ++env) {
  220. if (StringView { *env } == "_LOADER_BREAKPOINT=1") {
  221. g_do_breakpoint_trap_before_entry = true;
  222. }
  223. }
  224. }
  225. void ELF::DynamicLinker::linker_main(String&& main_program_name, int main_program_fd, bool is_secure, int argc, char** argv, char** envp)
  226. {
  227. g_envp = envp;
  228. g_allowed_to_check_environment_variables = !is_secure;
  229. if (g_allowed_to_check_environment_variables)
  230. read_environment_variables();
  231. map_library(main_program_name, main_program_fd);
  232. map_dependencies(main_program_name);
  233. dbgln_if(DYNAMIC_LOAD_DEBUG, "loaded all dependencies");
  234. for ([[maybe_unused]] auto& lib : g_loaders) {
  235. dbgln_if(DYNAMIC_LOAD_DEBUG, "{} - tls size: {}, tls offset: {}", lib.key, lib.value->tls_size(), lib.value->tls_offset());
  236. }
  237. allocate_tls();
  238. auto entry_point_function = [&main_program_name] {
  239. auto main_executable_loader = load_main_executable(main_program_name);
  240. auto entry_point = main_executable_loader->image().entry();
  241. if (main_executable_loader->is_dynamic())
  242. entry_point = entry_point.offset(main_executable_loader->base_address().get());
  243. return (EntryPointFunction)(entry_point.as_ptr());
  244. }();
  245. g_loaders.clear();
  246. int rc = syscall(SC_msyscall, nullptr);
  247. if (rc < 0) {
  248. VERIFY_NOT_REACHED();
  249. }
  250. dbgln_if(DYNAMIC_LOAD_DEBUG, "Jumping to entry point: {:p}", entry_point_function);
  251. if (g_do_breakpoint_trap_before_entry) {
  252. asm("int3");
  253. }
  254. rc = entry_point_function(argc, argv, envp);
  255. dbgln_if(DYNAMIC_LOAD_DEBUG, "rc: {}", rc);
  256. if (g_libc_exit != nullptr) {
  257. g_libc_exit(rc);
  258. } else {
  259. _exit(rc);
  260. }
  261. VERIFY_NOT_REACHED();
  262. }
  263. }