123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513 |
- /*
- * Copyright (c) 2020, Itamar S. <itamar8910@gmail.com>
- * Copyright (c) 2021, Andreas Kling <kling@serenityos.org>
- * Copyright (c) 2021, the SerenityOS developers.
- *
- * SPDX-License-Identifier: BSD-2-Clause
- */
- #include <AK/ByteBuffer.h>
- #include <AK/Debug.h>
- #include <AK/HashMap.h>
- #include <AK/HashTable.h>
- #include <AK/LexicalPath.h>
- #include <AK/NonnullRefPtrVector.h>
- #include <AK/ScopeGuard.h>
- #include <AK/Vector.h>
- #include <LibC/bits/pthread_integration.h>
- #include <LibC/link.h>
- #include <LibC/mman.h>
- #include <LibC/unistd.h>
- #include <LibDl/dlfcn.h>
- #include <LibDl/dlfcn_integration.h>
- #include <LibELF/AuxiliaryVector.h>
- #include <LibELF/DynamicLinker.h>
- #include <LibELF/DynamicLoader.h>
- #include <LibELF/DynamicObject.h>
- #include <LibELF/Hashes.h>
- #include <fcntl.h>
- #include <string.h>
- #include <sys/types.h>
- #include <syscall.h>
- namespace ELF {
- static HashMap<String, NonnullRefPtr<ELF::DynamicLoader>> s_loaders;
- static String s_main_program_name;
- static HashMap<String, NonnullRefPtr<ELF::DynamicObject>> s_global_objects;
- using EntryPointFunction = int (*)(int, char**, char**);
- using LibCExitFunction = void (*)(int);
- using DlIteratePhdrCallbackFunction = int (*)(struct dl_phdr_info*, size_t, void*);
- using DlIteratePhdrFunction = int (*)(DlIteratePhdrCallbackFunction, void*);
- static size_t s_current_tls_offset = 0;
- static size_t s_total_tls_size = 0;
- static size_t s_allocated_tls_block_size = 0;
- static char** s_envp = nullptr;
- static LibCExitFunction s_libc_exit = nullptr;
- static __pthread_mutex_t s_loader_lock = __PTHREAD_MUTEX_INITIALIZER;
- static bool s_allowed_to_check_environment_variables { false };
- static bool s_do_breakpoint_trap_before_entry { false };
- static Result<void, DlErrorMessage> __dlclose(void* handle);
- static Result<void*, DlErrorMessage> __dlopen(const char* filename, int flags);
- static Result<void*, DlErrorMessage> __dlsym(void* handle, const char* symbol_name);
- Optional<DynamicObject::SymbolLookupResult> DynamicLinker::lookup_global_symbol(const StringView& name)
- {
- Optional<DynamicObject::SymbolLookupResult> weak_result;
- auto symbol = DynamicObject::HashSymbol { name };
- for (auto& lib : s_global_objects) {
- auto res = lib.value->lookup_symbol(symbol);
- if (!res.has_value())
- continue;
- if (res.value().bind == STB_GLOBAL)
- return res;
- if (res.value().bind == STB_WEAK && !weak_result.has_value())
- weak_result = res;
- // We don't want to allow local symbols to be pulled in to other modules
- }
- return weak_result;
- }
- static String get_library_name(String path)
- {
- return LexicalPath(move(path)).basename();
- }
- static Result<NonnullRefPtr<DynamicLoader>, DlErrorMessage> map_library(const String& filename, int fd)
- {
- auto result = ELF::DynamicLoader::try_create(fd, filename);
- if (result.is_error()) {
- return result;
- }
- auto& loader = result.value();
- s_loaders.set(get_library_name(filename), *loader);
- loader->set_tls_offset(s_current_tls_offset);
- s_current_tls_offset += loader->tls_size_of_current_object();
- return loader;
- }
- static Result<NonnullRefPtr<DynamicLoader>, DlErrorMessage> map_library(const String& name)
- {
- if (name.contains("/")) {
- int fd = open(name.characters(), O_RDONLY);
- if (fd < 0)
- return DlErrorMessage { String::formatted("Could not open shared library: {}", name) };
- return map_library(name, fd);
- }
- // TODO: Do we want to also look for libs in other paths too?
- const char* search_paths[] = { "/usr/lib/{}", "/usr/local/lib/{}" };
- for (auto& search_path : search_paths) {
- auto path = String::formatted(search_path, name);
- int fd = open(path.characters(), O_RDONLY);
- if (fd < 0)
- continue;
- return map_library(name, fd);
- }
- return DlErrorMessage { String::formatted("Could not find required shared library: {}", name) };
- }
- static Vector<String> get_dependencies(const String& name)
- {
- auto lib = s_loaders.get(name).value();
- Vector<String> dependencies;
- lib->for_each_needed_library([&dependencies, &name](auto needed_name) {
- if (name == needed_name)
- return IterationDecision::Continue;
- dependencies.append(needed_name);
- return IterationDecision::Continue;
- });
- return dependencies;
- }
- static Result<void, DlErrorMessage> map_dependencies(const String& name)
- {
- dbgln_if(DYNAMIC_LOAD_DEBUG, "mapping dependencies for: {}", name);
- for (const auto& needed_name : get_dependencies(name)) {
- dbgln_if(DYNAMIC_LOAD_DEBUG, "needed library: {}", needed_name.characters());
- String library_name = get_library_name(needed_name);
- if (!s_loaders.contains(library_name) && !s_global_objects.contains(library_name)) {
- auto result1 = map_library(needed_name);
- if (result1.is_error()) {
- return result1.error();
- }
- auto result2 = map_dependencies(library_name);
- if (result2.is_error()) {
- return result2.error();
- }
- }
- }
- dbgln_if(DYNAMIC_LOAD_DEBUG, "mapped dependencies for {}", name);
- return {};
- }
- static void allocate_tls()
- {
- s_total_tls_size = 0;
- for (const auto& data : s_loaders) {
- dbgln_if(DYNAMIC_LOAD_DEBUG, "{}: TLS Size: {}", data.key, data.value->tls_size_of_current_object());
- s_total_tls_size += data.value->tls_size_of_current_object();
- }
- if (!s_total_tls_size)
- return;
- auto page_aligned_size = align_up_to(s_total_tls_size, PAGE_SIZE);
- ByteBuffer initial_tls_data = ByteBuffer::create_zeroed(page_aligned_size);
- // Initialize TLS data
- for (const auto& entry : s_loaders) {
- entry.value->copy_initial_tls_data_into(initial_tls_data);
- }
- void* master_tls = ::allocate_tls((char*)initial_tls_data.data(), initial_tls_data.size());
- VERIFY(master_tls != (void*)-1);
- dbgln_if(DYNAMIC_LOAD_DEBUG, "from userspace, master_tls: {:p}", master_tls);
- s_allocated_tls_block_size = initial_tls_data.size();
- }
- static int __dl_iterate_phdr(DlIteratePhdrCallbackFunction callback, void* data)
- {
- __pthread_mutex_lock(&s_loader_lock);
- ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
- for (auto& it : s_global_objects) {
- auto& object = it.value;
- auto info = dl_phdr_info {
- .dlpi_addr = (ElfW(Addr))object->base_address().as_ptr(),
- .dlpi_name = object->filename().characters(),
- .dlpi_phdr = object->program_headers(),
- .dlpi_phnum = object->program_header_count()
- };
- auto res = callback(&info, sizeof(info), data);
- if (res != 0)
- return res;
- }
- return 0;
- }
- static void initialize_libc(DynamicObject& libc)
- {
- // Traditionally, `_start` of the main program initializes libc.
- // However, since some libs use malloc() and getenv() in global constructors,
- // we have to initialize libc just after it is loaded.
- // Also, we can't just mark `__libc_init` with "__attribute__((constructor))"
- // because it uses getenv() internally, so `environ` has to be initialized before we call `__libc_init`.
- auto res = libc.lookup_symbol("environ"sv);
- VERIFY(res.has_value());
- *((char***)res.value().address.as_ptr()) = s_envp;
- res = libc.lookup_symbol("__environ_is_malloced"sv);
- VERIFY(res.has_value());
- *((bool*)res.value().address.as_ptr()) = false;
- res = libc.lookup_symbol("exit"sv);
- VERIFY(res.has_value());
- s_libc_exit = (LibCExitFunction)res.value().address.as_ptr();
- res = libc.lookup_symbol("__dl_iterate_phdr"sv);
- VERIFY(res.has_value());
- *((DlIteratePhdrFunction*)res.value().address.as_ptr()) = __dl_iterate_phdr;
- res = libc.lookup_symbol("__dlclose"sv);
- VERIFY(res.has_value());
- *((DlCloseFunction*)res.value().address.as_ptr()) = __dlclose;
- res = libc.lookup_symbol("__dlopen"sv);
- VERIFY(res.has_value());
- *((DlOpenFunction*)res.value().address.as_ptr()) = __dlopen;
- res = libc.lookup_symbol("__dlsym"sv);
- VERIFY(res.has_value());
- *((DlSymFunction*)res.value().address.as_ptr()) = __dlsym;
- res = libc.lookup_symbol("__libc_init"sv);
- VERIFY(res.has_value());
- typedef void libc_init_func();
- ((libc_init_func*)res.value().address.as_ptr())();
- }
- template<typename Callback>
- static void for_each_unfinished_dependency_of(const String& name, HashTable<String>& seen_names, Callback callback)
- {
- if (!s_loaders.contains(name))
- return;
- if (seen_names.contains(name))
- return;
- seen_names.set(name);
- for (const auto& needed_name : get_dependencies(name))
- for_each_unfinished_dependency_of(get_library_name(needed_name), seen_names, callback);
- callback(*s_loaders.get(name).value());
- }
- static NonnullRefPtrVector<DynamicLoader> collect_loaders_for_library(const String& name)
- {
- HashTable<String> seen_names;
- NonnullRefPtrVector<DynamicLoader> loaders;
- for_each_unfinished_dependency_of(name, seen_names, [&](auto& loader) {
- loaders.append(loader);
- });
- return loaders;
- }
- static Result<NonnullRefPtr<DynamicLoader>, DlErrorMessage> load_main_library(const String& name, int flags)
- {
- auto main_library_loader = *s_loaders.get(name);
- auto main_library_object = main_library_loader->map();
- s_global_objects.set(name, *main_library_object);
- auto loaders = collect_loaders_for_library(name);
- for (auto& loader : loaders) {
- auto dynamic_object = loader.map();
- if (dynamic_object)
- s_global_objects.set(dynamic_object->filename(), *dynamic_object);
- }
- for (auto& loader : loaders) {
- bool success = loader.link(flags);
- if (!success) {
- return DlErrorMessage { String::formatted("Failed to link library {}", loader.filename()) };
- }
- }
- for (auto& loader : loaders) {
- auto result = loader.load_stage_3(flags);
- VERIFY(!result.is_error());
- auto& object = result.value();
- if (loader.filename() == "libsystem.so") {
- if (syscall(SC_msyscall, object->base_address().as_ptr())) {
- VERIFY_NOT_REACHED();
- }
- }
- if (loader.filename() == "libc.so") {
- initialize_libc(*object);
- }
- }
- for (auto& loader : loaders) {
- loader.load_stage_4();
- }
- return main_library_loader;
- }
- static Result<void, DlErrorMessage> __dlclose(void* handle)
- {
- dbgln_if(DYNAMIC_LOAD_DEBUG, "__dlclose: {}", handle);
- __pthread_mutex_lock(&s_loader_lock);
- ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
- // FIXME: this will not currently destroy the dynamic object
- // because we're intentionally holding a strong reference to it
- // via s_global_objects until there's proper unload support.
- auto object = static_cast<ELF::DynamicObject*>(handle);
- object->unref();
- return {};
- }
- static Optional<DlErrorMessage> verify_tls_for_dlopen(const DynamicLoader& loader)
- {
- if (loader.tls_size_of_current_object() == 0)
- return {};
- if (s_total_tls_size + loader.tls_size_of_current_object() > s_allocated_tls_block_size)
- return DlErrorMessage("TLS size too large");
- bool tls_data_is_all_zero = true;
- loader.image().for_each_program_header([&loader, &tls_data_is_all_zero](ELF::Image::ProgramHeader program_header) {
- if (program_header.type() != PT_TLS)
- return IterationDecision::Continue;
- auto* tls_data = (const u8*)loader.image().base_address() + program_header.offset();
- for (size_t i = 0; i < program_header.size_in_image(); ++i) {
- if (tls_data[i] != 0) {
- tls_data_is_all_zero = false;
- break;
- }
- }
- return IterationDecision::Break;
- });
- if (tls_data_is_all_zero)
- return {};
- return DlErrorMessage("Using dlopen() with libraries that have non-zeroed TLS is currently not supported");
- }
- static Result<void*, DlErrorMessage> __dlopen(const char* filename, int flags)
- {
- // FIXME: RTLD_NOW and RTLD_LOCAL are not supported
- flags &= ~RTLD_NOW;
- flags |= RTLD_LAZY;
- flags &= ~RTLD_LOCAL;
- flags |= RTLD_GLOBAL;
- dbgln_if(DYNAMIC_LOAD_DEBUG, "__dlopen invoked, filename={}, flags={}", filename, flags);
- auto library_name = get_library_name(filename ? filename : s_main_program_name);
- if (__pthread_mutex_trylock(&s_loader_lock) != 0)
- return DlErrorMessage { "Nested calls to dlopen() are not permitted." };
- ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
- auto existing_elf_object = s_global_objects.get(library_name);
- if (existing_elf_object.has_value()) {
- // It's up to the caller to release the ref with dlclose().
- return &existing_elf_object->leak_ref();
- }
- VERIFY(!library_name.is_empty());
- auto result1 = map_library(filename);
- if (result1.is_error()) {
- return result1.error();
- }
- if (auto error = verify_tls_for_dlopen(result1.value()); error.has_value())
- return error.value();
- auto result2 = map_dependencies(library_name);
- if (result2.is_error()) {
- return result2.error();
- }
- auto result = load_main_library(library_name, flags);
- if (result.is_error())
- return result.error();
- s_total_tls_size += result1.value()->tls_size_of_current_object();
- auto object = s_global_objects.get(library_name);
- if (!object.has_value())
- return DlErrorMessage { "Could not load ELF object." };
- // It's up to the caller to release the ref with dlclose().
- return &object->leak_ref();
- }
- static Result<void*, DlErrorMessage> __dlsym(void* handle, const char* symbol_name)
- {
- dbgln_if(DYNAMIC_LOAD_DEBUG, "__dlsym: {}, {}", handle, symbol_name);
- __pthread_mutex_lock(&s_loader_lock);
- ScopeGuard unlock_guard = [] { __pthread_mutex_unlock(&s_loader_lock); };
- auto object = static_cast<DynamicObject*>(handle);
- if (!handle) {
- auto library_name = get_library_name(s_main_program_name);
- auto global_object = s_global_objects.get(library_name);
- object = *global_object;
- }
- auto symbol = object->lookup_symbol(symbol_name);
- if (!symbol.has_value()) {
- return DlErrorMessage { String::formatted("Symbol {} not found", symbol_name) };
- }
- return symbol.value().address.as_ptr();
- }
- static void read_environment_variables()
- {
- for (char** env = s_envp; *env; ++env) {
- if (StringView { *env } == "_LOADER_BREAKPOINT=1") {
- s_do_breakpoint_trap_before_entry = true;
- }
- }
- }
- void ELF::DynamicLinker::linker_main(String&& main_program_name, int main_program_fd, bool is_secure, int argc, char** argv, char** envp)
- {
- s_envp = envp;
- s_allowed_to_check_environment_variables = !is_secure;
- if (s_allowed_to_check_environment_variables)
- read_environment_variables();
- s_main_program_name = main_program_name;
- auto library_name = get_library_name(main_program_name);
- // NOTE: We always map the main library first, since it may require
- // placement at a specific address.
- auto result1 = map_library(main_program_name, main_program_fd);
- if (result1.is_error()) {
- warnln("{}", result1.error().text);
- fflush(stderr);
- _exit(1);
- }
- result1.release_value();
- auto result2 = map_dependencies(library_name);
- if (result2.is_error()) {
- warnln("{}", result2.error().text);
- fflush(stderr);
- _exit(1);
- }
- dbgln_if(DYNAMIC_LOAD_DEBUG, "loaded all dependencies");
- for ([[maybe_unused]] auto& lib : s_loaders) {
- dbgln_if(DYNAMIC_LOAD_DEBUG, "{} - tls size: {}, tls offset: {}", lib.key, lib.value->tls_size_of_current_object(), lib.value->tls_offset());
- }
- allocate_tls();
- auto entry_point_function = [&main_program_name] {
- auto library_name = get_library_name(main_program_name);
- auto result = load_main_library(library_name, RTLD_GLOBAL | RTLD_LAZY);
- if (result.is_error()) {
- warnln("{}", result.error().text);
- _exit(1);
- }
- auto& main_executable_loader = result.value();
- auto entry_point = main_executable_loader->image().entry();
- if (main_executable_loader->is_dynamic())
- entry_point = entry_point.offset(main_executable_loader->base_address().get());
- return (EntryPointFunction)(entry_point.as_ptr());
- }();
- s_loaders.clear();
- int rc = syscall(SC_msyscall, nullptr);
- if (rc < 0) {
- VERIFY_NOT_REACHED();
- }
- dbgln_if(DYNAMIC_LOAD_DEBUG, "Jumping to entry point: {:p}", entry_point_function);
- if (s_do_breakpoint_trap_before_entry) {
- asm("int3");
- }
- rc = entry_point_function(argc, argv, envp);
- dbgln_if(DYNAMIC_LOAD_DEBUG, "rc: {}", rc);
- if (s_libc_exit != nullptr) {
- s_libc_exit(rc);
- } else {
- _exit(rc);
- }
- VERIFY_NOT_REACHED();
- }
- }
|