/* * Copyright (c) 2018-2023, Andreas Kling * * SPDX-License-Identifier: BSD-2-Clause */ #pragma once #include #include #include #include #include #include #include #include #include #include #include #include #ifdef ENABLE_KERNEL_COVERAGE_COLLECTION # include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include namespace Kernel { MutexProtected>& hostname(); UnixDateTime kgettimeofday(); #define ENUMERATE_PLEDGE_PROMISES \ __ENUMERATE_PLEDGE_PROMISE(stdio) \ __ENUMERATE_PLEDGE_PROMISE(rpath) \ __ENUMERATE_PLEDGE_PROMISE(wpath) \ __ENUMERATE_PLEDGE_PROMISE(cpath) \ __ENUMERATE_PLEDGE_PROMISE(dpath) \ __ENUMERATE_PLEDGE_PROMISE(inet) \ __ENUMERATE_PLEDGE_PROMISE(id) \ __ENUMERATE_PLEDGE_PROMISE(proc) \ __ENUMERATE_PLEDGE_PROMISE(ptrace) \ __ENUMERATE_PLEDGE_PROMISE(exec) \ __ENUMERATE_PLEDGE_PROMISE(unix) \ __ENUMERATE_PLEDGE_PROMISE(recvfd) \ __ENUMERATE_PLEDGE_PROMISE(sendfd) \ __ENUMERATE_PLEDGE_PROMISE(fattr) \ __ENUMERATE_PLEDGE_PROMISE(tty) \ __ENUMERATE_PLEDGE_PROMISE(chown) \ __ENUMERATE_PLEDGE_PROMISE(thread) \ __ENUMERATE_PLEDGE_PROMISE(video) \ __ENUMERATE_PLEDGE_PROMISE(accept) \ __ENUMERATE_PLEDGE_PROMISE(settime) \ __ENUMERATE_PLEDGE_PROMISE(sigaction) \ __ENUMERATE_PLEDGE_PROMISE(setkeymap) \ __ENUMERATE_PLEDGE_PROMISE(prot_exec) \ __ENUMERATE_PLEDGE_PROMISE(map_fixed) \ __ENUMERATE_PLEDGE_PROMISE(getkeymap) \ __ENUMERATE_PLEDGE_PROMISE(jail) \ __ENUMERATE_PLEDGE_PROMISE(mount) \ __ENUMERATE_PLEDGE_PROMISE(no_error) #define __ENUMERATE_PLEDGE_PROMISE(x) sizeof(#x) + 1 + // NOTE: We truncate the last space from the string as it's not needed (with 0 - 1). constexpr static unsigned all_promises_strings_length_with_spaces = ENUMERATE_PLEDGE_PROMISES 0 - 1; #undef __ENUMERATE_PLEDGE_PROMISE // NOTE: This is a sanity check because length of more than 1024 characters // is not reasonable. static_assert(all_promises_strings_length_with_spaces <= 1024); enum class Pledge : u32 { #define __ENUMERATE_PLEDGE_PROMISE(x) x, ENUMERATE_PLEDGE_PROMISES #undef __ENUMERATE_PLEDGE_PROMISE }; enum class VeilState { None, Dropped, Locked, LockedInherited, }; static constexpr FlatPtr futex_key_private_flag = 0b1; union GlobalFutexKey { struct { Memory::VMObject const* vmobject; FlatPtr offset; } shared; struct { Memory::AddressSpace const* address_space; FlatPtr user_address; } private_; struct { FlatPtr parent; FlatPtr offset; } raw; }; static_assert(sizeof(GlobalFutexKey) == (sizeof(FlatPtr) * 2)); struct LoadResult; class ProcessList; class Process final : public ListedRefCounted , public LockWeakable { class ProtectedValues { public: ProcessID pid { 0 }; ProcessID ppid { 0 }; // FIXME: This should be a NonnullRefPtr RefPtr credentials; RefPtr process_group; RefPtr tty; bool dumpable { false }; bool executable_is_setid { false }; bool has_promises { false }; u32 promises { 0 }; bool has_execpromises { false }; u32 execpromises { 0 }; mode_t umask { 022 }; VirtualAddress signal_trampoline; Atomic thread_count { 0 }; u8 termination_status { 0 }; u8 termination_signal { 0 }; SetOnce reject_transition_to_executable_from_writable_prot; }; public: AK_MAKE_NONCOPYABLE(Process); AK_MAKE_NONMOVABLE(Process); MAKE_ALIGNED_ALLOCATED(Process, PAGE_SIZE); friend class Thread; friend class Coredump; auto with_protected_data(auto&& callback) const { SpinlockLocker locker(m_protected_data_lock); return callback(m_protected_values_do_not_access_directly); } auto with_mutable_protected_data(auto&& callback) { SpinlockLocker locker(m_protected_data_lock); unprotect_data(); auto guard = ScopeGuard([&] { protect_data(); }); return callback(m_protected_values_do_not_access_directly); } enum class State : u8 { Running = 0, Dying, Dead }; public: static Process& current() { auto* current_thread = Processor::current_thread(); VERIFY(current_thread); return current_thread->process(); } static bool has_current() { return Processor::current_thread() != nullptr; } template static void kernel_process_trampoline(void* data) { EntryFunction* func = reinterpret_cast(data); (*func)(); delete func; } enum class RegisterProcess { No, Yes }; struct ProcessAndFirstThread { NonnullRefPtr process; NonnullRefPtr first_thread; }; template static ErrorOr create_kernel_process(StringView name, EntryFunction entry, u32 affinity = THREAD_AFFINITY_DEFAULT, RegisterProcess do_register = RegisterProcess::Yes) { auto* entry_func = new EntryFunction(move(entry)); return create_kernel_process(name, &Process::kernel_process_trampoline, entry_func, affinity, do_register); } static ErrorOr create_kernel_process(StringView name, void (*entry)(void*), void* entry_data = nullptr, u32 affinity = THREAD_AFFINITY_DEFAULT, RegisterProcess do_register = RegisterProcess::Yes); static ErrorOr create_user_process(StringView path, UserID, GroupID, Vector> arguments, Vector> environment, RefPtr); static void register_new(Process&); ~Process(); virtual void remove_from_secondary_lists(); ErrorOr> create_kernel_thread(void (*entry)(void*), void* entry_data, u32 priority, StringView name, u32 affinity = THREAD_AFFINITY_DEFAULT, bool joinable = true); bool is_profiling() const { return m_profiling; } void set_profiling(bool profiling) { m_profiling = profiling; } #ifdef ENABLE_KERNEL_COVERAGE_COLLECTION NO_SANITIZE_COVERAGE KCOVInstance* kcov_instance() { return m_kcov_instance; } void set_kcov_instance(KCOVInstance* kcov_instance) { m_kcov_instance = kcov_instance; } static bool is_kcov_busy(); #endif bool should_generate_coredump() const { return m_should_generate_coredump; } void set_should_generate_coredump(bool b) { m_should_generate_coredump = b; } bool is_dying() const { return m_state.load(AK::MemoryOrder::memory_order_acquire) != State::Running; } bool is_dead() const { return m_state.load(AK::MemoryOrder::memory_order_acquire) == State::Dead; } bool is_stopped() const { return m_is_stopped; } bool set_stopped(bool stopped) { return m_is_stopped.exchange(stopped); } bool is_kernel_process() const { return m_is_kernel_process; } bool is_user_process() const { return !m_is_kernel_process; } static RefPtr from_pid_in_same_jail(ProcessID); static RefPtr from_pid_ignoring_jails(ProcessID); static SessionID get_sid_from_pgid(ProcessGroupID pgid); using Name = FixedStringBuffer<32>; SpinlockProtected const& name() const; void set_name(StringView); ProcessID pid() const { return with_protected_data([](auto& protected_data) { return protected_data.pid; }); } SessionID sid() const { return credentials()->sid(); } bool is_session_leader() const { return sid().value() == pid().value(); } ProcessGroupID pgid() const { return with_protected_data([](auto& protected_data) { return protected_data.process_group ? protected_data.process_group->pgid() : 0; }); } bool is_group_leader() const { return pgid().value() == pid().value(); } ProcessID ppid() const { return with_protected_data([](auto& protected_data) { return protected_data.ppid; }); } SpinlockProtected, LockRank::Process> const& jail() { return m_attached_jail; } bool is_currently_in_jail() const { return m_attached_jail.with([&](auto& jail) -> bool { return !jail.is_null(); }); } NonnullRefPtr credentials() const; bool is_dumpable() const { return with_protected_data([](auto& protected_data) { return protected_data.dumpable; }); } mode_t umask() const { return with_protected_data([](auto& protected_data) { return protected_data.umask; }); } // Breakable iteration functions template Callback> static void for_each_ignoring_jails(Callback); static ErrorOr for_each_in_same_jail(Function(Process&)>); ErrorOr for_each_in_pgrp_in_same_jail(ProcessGroupID, Function(Process&)>); ErrorOr for_each_child_in_same_jail(Function(Process&)>); template Callback> IterationDecision for_each_thread(Callback); template Callback> IterationDecision for_each_thread(Callback callback) const; ErrorOr try_for_each_thread(Function(Thread const&)>) const; // Non-breakable iteration functions template Callback> static void for_each_ignoring_jails(Callback); template Callback> IterationDecision for_each_thread(Callback); template Callback> IterationDecision for_each_thread(Callback callback) const; void die(); void finalize(); ThreadTracer* tracer() { return m_tracer.ptr(); } bool is_traced() const { return !!m_tracer; } ErrorOr start_tracing_from(ProcessID tracer); void stop_tracing(); void tracer_trap(Thread&, RegisterState const&); ErrorOr sys$emuctl(); ErrorOr sys$yield(); ErrorOr sys$sync(); ErrorOr sys$beep(int tone); ErrorOr sys$create_inode_watcher(u32 flags); ErrorOr sys$inode_watcher_add_watch(Userspace user_params); ErrorOr sys$inode_watcher_remove_watch(int fd, int wd); ErrorOr sys$dbgputstr(Userspace, size_t); ErrorOr sys$dump_backtrace(); ErrorOr sys$gettid(); ErrorOr sys$setsid(); ErrorOr sys$getsid(pid_t); ErrorOr sys$setpgid(pid_t pid, pid_t pgid); ErrorOr sys$getpgrp(); ErrorOr sys$getpgid(pid_t); ErrorOr sys$getuid(); ErrorOr sys$getgid(); ErrorOr sys$geteuid(); ErrorOr sys$getegid(); ErrorOr sys$getpid(); ErrorOr sys$getppid(); ErrorOr sys$getresuid(Userspace, Userspace, Userspace); ErrorOr sys$getresgid(Userspace, Userspace, Userspace); ErrorOr sys$getrusage(int, Userspace); ErrorOr sys$umask(mode_t); ErrorOr sys$open(Userspace); ErrorOr sys$close(int fd); ErrorOr sys$read(int fd, Userspace, size_t); ErrorOr sys$pread(int fd, Userspace, size_t, off_t); ErrorOr sys$readv(int fd, Userspace iov, int iov_count); ErrorOr sys$write(int fd, Userspace, size_t); ErrorOr sys$pwritev(int fd, Userspace iov, int iov_count, off_t); ErrorOr sys$fstat(int fd, Userspace); ErrorOr sys$stat(Userspace); ErrorOr sys$annotate_mapping(Userspace, int flags); ErrorOr sys$lseek(int fd, Userspace, int whence); ErrorOr sys$ftruncate(int fd, off_t); ErrorOr sys$futimens(Userspace); ErrorOr sys$posix_fallocate(int fd, off_t, off_t); ErrorOr sys$kill(pid_t pid_or_pgid, int sig); [[noreturn]] void sys$exit(int status); ErrorOr sys$sigreturn(RegisterState& registers); ErrorOr sys$waitid(Userspace); ErrorOr sys$mmap(Userspace); ErrorOr sys$mremap(Userspace); ErrorOr sys$munmap(Userspace, size_t); ErrorOr sys$set_mmap_name(Userspace); ErrorOr sys$mprotect(Userspace, size_t, int prot); ErrorOr sys$madvise(Userspace, size_t, int advice); ErrorOr sys$msync(Userspace, size_t, int flags); ErrorOr sys$purge(int mode); ErrorOr sys$poll(Userspace); ErrorOr sys$get_dir_entries(int fd, Userspace, size_t); ErrorOr sys$getcwd(Userspace, size_t); ErrorOr sys$chdir(Userspace, size_t); ErrorOr sys$fchdir(int fd); ErrorOr sys$adjtime(Userspace, Userspace); ErrorOr sys$clock_gettime(clockid_t, Userspace); ErrorOr sys$clock_settime(clockid_t, Userspace); ErrorOr sys$clock_nanosleep(Userspace); ErrorOr sys$clock_getres(Userspace); ErrorOr sys$gethostname(Userspace, size_t); ErrorOr sys$sethostname(Userspace, size_t); ErrorOr sys$uname(Userspace); ErrorOr sys$readlink(Userspace); ErrorOr sys$fork(RegisterState&); ErrorOr sys$execve(Userspace); ErrorOr sys$dup2(int old_fd, int new_fd); ErrorOr sys$sigaction(int signum, Userspace act, Userspace old_act); ErrorOr sys$sigaltstack(Userspace ss, Userspace old_ss); ErrorOr sys$sigprocmask(int how, Userspace set, Userspace old_set); ErrorOr sys$sigpending(Userspace); ErrorOr sys$sigsuspend(Userspace); ErrorOr sys$sigtimedwait(Userspace, Userspace, Userspace); ErrorOr sys$getgroups(size_t, Userspace); ErrorOr sys$setgroups(size_t, Userspace); ErrorOr sys$pipe(Userspace, int flags); ErrorOr sys$killpg(pid_t pgrp, int sig); ErrorOr sys$seteuid(UserID); ErrorOr sys$setegid(GroupID); ErrorOr sys$setuid(UserID); ErrorOr sys$setgid(GroupID); ErrorOr sys$setreuid(UserID, UserID); ErrorOr sys$setresuid(UserID, UserID, UserID); ErrorOr sys$setregid(GroupID, GroupID); ErrorOr sys$setresgid(GroupID, GroupID, GroupID); ErrorOr sys$alarm(unsigned seconds); ErrorOr sys$faccessat(Userspace); ErrorOr sys$fcntl(int fd, int cmd, uintptr_t extra_arg); ErrorOr sys$ioctl(int fd, unsigned request, FlatPtr arg); ErrorOr sys$mkdir(int dirfd, Userspace pathname, size_t path_length, mode_t mode); ErrorOr sys$times(Userspace); ErrorOr sys$utime(Userspace pathname, size_t path_length, Userspace); ErrorOr sys$utimensat(Userspace); ErrorOr sys$link(Userspace); ErrorOr sys$unlink(int dirfd, Userspace pathname, size_t path_length, int flags); ErrorOr sys$symlink(Userspace); ErrorOr sys$rmdir(Userspace pathname, size_t path_length); ErrorOr sys$fsmount(Userspace); ErrorOr sys$fsopen(Userspace); ErrorOr sys$umount(Userspace mountpoint, size_t mountpoint_length); ErrorOr sys$chmod(Userspace); ErrorOr sys$fchmod(int fd, mode_t); ErrorOr sys$chown(Userspace); ErrorOr sys$fchown(int fd, UserID, GroupID); ErrorOr sys$fsync(int fd); ErrorOr sys$socket(int domain, int type, int protocol); ErrorOr sys$bind(int sockfd, Userspace addr, socklen_t); ErrorOr sys$listen(int sockfd, int backlog); ErrorOr sys$accept4(Userspace); ErrorOr sys$connect(int sockfd, Userspace, socklen_t); ErrorOr sys$shutdown(int sockfd, int how); ErrorOr sys$sendmsg(int sockfd, Userspace, int flags); ErrorOr sys$recvmsg(int sockfd, Userspace, int flags); ErrorOr sys$getsockopt(Userspace); ErrorOr sys$setsockopt(Userspace); ErrorOr sys$getsockname(Userspace); ErrorOr sys$getpeername(Userspace); ErrorOr sys$socketpair(Userspace); ErrorOr sys$scheduler_set_parameters(Userspace); ErrorOr sys$scheduler_get_parameters(Userspace); ErrorOr sys$create_thread(void* (*)(void*), Userspace); [[noreturn]] void sys$exit_thread(Userspace, Userspace, size_t); ErrorOr sys$join_thread(pid_t tid, Userspace exit_value); ErrorOr sys$detach_thread(pid_t tid); ErrorOr sys$kill_thread(pid_t tid, int signal); ErrorOr sys$rename(Userspace); ErrorOr sys$mknod(Userspace); ErrorOr sys$realpath(Userspace); ErrorOr sys$getrandom(Userspace, size_t, unsigned int); ErrorOr sys$getkeymap(Userspace); ErrorOr sys$setkeymap(Userspace); ErrorOr sys$profiling_enable(pid_t, u64); ErrorOr profiling_enable(pid_t, u64 event_mask); ErrorOr sys$profiling_disable(pid_t); ErrorOr sys$profiling_free_buffer(pid_t); ErrorOr sys$futex(Userspace); ErrorOr sys$pledge(Userspace); ErrorOr sys$unveil(Userspace); ErrorOr sys$perf_event(int type, FlatPtr arg1, FlatPtr arg2); ErrorOr sys$perf_register_string(Userspace, size_t); ErrorOr sys$get_stack_bounds(Userspace stack_base, Userspace stack_size); ErrorOr sys$ptrace(Userspace); ErrorOr sys$sendfd(int sockfd, int fd); ErrorOr sys$recvfd(int sockfd, int options); ErrorOr sys$sysconf(int name); ErrorOr sys$disown(ProcessID); ErrorOr sys$prctl(int option, FlatPtr arg1, FlatPtr arg2, FlatPtr arg3); ErrorOr sys$anon_create(size_t, int options); ErrorOr sys$statvfs(Userspace user_params); ErrorOr sys$fstatvfs(int fd, statvfs* buf); ErrorOr sys$map_time_page(); ErrorOr sys$jail_create(Userspace user_params); ErrorOr sys$jail_attach(Userspace user_params); ErrorOr sys$get_root_session_id(pid_t force_sid); ErrorOr sys$remount(Userspace user_params); ErrorOr sys$bindmount(Userspace user_params); ErrorOr sys$archctl(int option, FlatPtr arg1); enum SockOrPeerName { SockName, PeerName, }; template ErrorOr get_sock_or_peer_name(Params const&); static void initialize(); [[noreturn]] void crash(int signal, Optional regs, bool out_of_memory = false); [[nodiscard]] siginfo_t wait_info() const; RefPtr tty(); RefPtr tty() const; void set_tty(RefPtr); clock_t m_ticks_in_user { 0 }; clock_t m_ticks_in_kernel { 0 }; clock_t m_ticks_in_user_for_dead_children { 0 }; clock_t m_ticks_in_kernel_for_dead_children { 0 }; NonnullRefPtr current_directory(); RefPtr executable(); RefPtr executable() const; UnixDateTime creation_time() const { return m_creation_time; } static constexpr size_t max_arguments_size = Thread::default_userspace_stack_size / 8; static constexpr size_t max_environment_size = Thread::default_userspace_stack_size / 8; static constexpr size_t max_auxiliary_size = Thread::default_userspace_stack_size / 8; Vector> const& arguments() const { return m_arguments; } Vector> const& environment() const { return m_environment; } ErrorOr exec(NonnullOwnPtr path, Vector> arguments, Vector> environment, Thread*& new_main_thread, InterruptsState& previous_interrupts_state, int recursion_depth = 0); ErrorOr load(Memory::AddressSpace& new_space, NonnullRefPtr main_program_description, RefPtr interpreter_description, Elf_Ehdr const& main_program_header, Optional minimum_stack_size = {}); void terminate_due_to_signal(u8 signal); ErrorOr send_signal(u8 signal, Process* sender); u8 termination_signal() const { return with_protected_data([](auto& protected_data) -> u8 { return protected_data.termination_signal; }); } u8 termination_status() const { return with_protected_data([](auto& protected_data) { return protected_data.termination_status; }); } u16 thread_count() const { return with_protected_data([](auto& protected_data) { return protected_data.thread_count.load(AK::MemoryOrder::memory_order_relaxed); }); } Mutex& big_lock() { return m_big_lock; } Mutex& ptrace_lock() { return m_ptrace_lock; } bool has_promises() const { return with_protected_data([](auto& protected_data) { return protected_data.has_promises; }); } bool has_promised(Pledge pledge) const { return with_protected_data([&](auto& protected_data) { return (protected_data.promises & (1U << (u32)pledge)) != 0; }); } VeilState veil_state() const { return m_unveil_data.with([&](auto const& unveil_data) { return unveil_data.state; }); } struct UnveilData { explicit UnveilData(UnveilNode&& p) : paths(move(p)) { } VeilState state { VeilState::None }; UnveilNode paths; }; auto& unveil_data() { return m_unveil_data; } auto const& unveil_data() const { return m_unveil_data; } auto& exec_unveil_data() { return m_exec_unveil_data; } auto const& exec_unveil_data() const { return m_exec_unveil_data; } bool wait_for_tracer_at_next_execve() const { return m_wait_for_tracer_at_next_execve; } void set_wait_for_tracer_at_next_execve(bool val) { m_wait_for_tracer_at_next_execve = val; } ErrorOr peek_user_data(Span destination, Userspace address); ErrorOr peek_user_data(Userspace address); ErrorOr poke_user_data(Userspace address, FlatPtr data); void disowned_by_waiter(Process& process); void unblock_waiters(Thread::WaitBlocker::UnblockFlags, u8 signal = 0); Thread::WaitBlockerSet& wait_blocker_set() { return m_wait_blocker_set; } template ErrorOr for_each_coredump_property(Callback callback) const { return m_coredump_properties.with([&](auto const& coredump_properties) -> ErrorOr { for (auto const& property : coredump_properties) { if (property.key && property.value) TRY(callback(*property.key, *property.value)); } return {}; }); } ErrorOr set_coredump_property(NonnullOwnPtr key, NonnullOwnPtr value); ErrorOr try_set_coredump_property(StringView key, StringView value); Vector> const& threads_for_coredump(Badge) const { return m_threads_for_coredump; } PerformanceEventBuffer* perf_events() { return m_perf_event_buffer; } PerformanceEventBuffer const* perf_events() const { return m_perf_event_buffer; } SpinlockProtected, LockRank::None>& address_space() { return m_space; } SpinlockProtected, LockRank::None> const& address_space() const { return m_space; } VirtualAddress signal_trampoline() const { return with_protected_data([](auto& protected_data) { return protected_data.signal_trampoline; }); } ErrorOr require_promise(Pledge); ErrorOr require_no_promises() const; bool should_reject_transition_to_executable_from_writable_prot() const { return with_protected_data([](auto& protected_data) { return protected_data.reject_transition_to_executable_from_writable_prot.was_set(); }); } ErrorOr validate_mmap_prot(int prot, bool map_stack, bool map_anonymous, Memory::Region const* region = nullptr) const; ErrorOr validate_inode_mmap_prot(int prot, bool description_readable, bool description_writable, bool map_shared) const; template static ErrorOr> get_syscall_string_fixed_buffer(Syscall::StringArgument const& argument) { // NOTE: If the string is too much big for the FixedStringBuffer, // we return E2BIG error here. FixedStringBuffer buffer; TRY(try_copy_string_from_user_into_fixed_string_buffer(reinterpret_cast(argument.characters), buffer, argument.length)); return buffer; } template static ErrorOr> get_syscall_name_string_fixed_buffer(Userspace user_buffer, size_t user_length = Size) { // NOTE: If the string is too much big for the FixedStringBuffer, // we return E2BIG error here. FixedStringBuffer buffer; TRY(try_copy_string_from_user_into_fixed_string_buffer(user_buffer, buffer, user_length)); return buffer; } template static ErrorOr> get_syscall_name_string_fixed_buffer(Syscall::StringArgument const& argument) { // NOTE: If the string is too much big for the FixedStringBuffer, // we return ENAMETOOLONG error here. FixedStringBuffer buffer; TRY(try_copy_name_from_user_into_fixed_string_buffer(reinterpret_cast(argument.characters), buffer, argument.length)); return buffer; } private: friend class MemoryManager; friend class Scheduler; friend class Region; friend class PerformanceManager; bool add_thread(Thread&); bool remove_thread(Thread&); Process(StringView name, NonnullRefPtr, ProcessID ppid, bool is_kernel_process, RefPtr current_directory, RefPtr executable, RefPtr tty, UnveilNode unveil_tree, UnveilNode exec_unveil_tree, UnixDateTime creation_time); static ErrorOr create_with_forked_name(UserID, GroupID, ProcessID ppid, bool is_kernel_process, RefPtr current_directory = nullptr, RefPtr executable = nullptr, RefPtr = nullptr, Process* fork_parent = nullptr); static ErrorOr create(StringView name, UserID, GroupID, ProcessID ppid, bool is_kernel_process, RefPtr current_directory = nullptr, RefPtr executable = nullptr, RefPtr = nullptr, Process* fork_parent = nullptr); ErrorOr> attach_resources(NonnullOwnPtr&&, Process* fork_parent); static ProcessID allocate_pid(); void kill_threads_except_self(); void kill_all_threads(); ErrorOr dump_core(); ErrorOr dump_perfcore(); bool create_perf_events_buffer_if_needed(); void delete_perf_events_buffer(); ErrorOr do_exec(NonnullRefPtr main_program_description, Vector> arguments, Vector> environment, RefPtr interpreter_description, Thread*& new_main_thread, InterruptsState& previous_interrupts_state, Elf_Ehdr const& main_program_header, Optional minimum_stack_size = {}); ErrorOr do_write(OpenFileDescription&, UserOrKernelBuffer const&, size_t, Optional = {}); ErrorOr do_statvfs(FileSystem const& path, Custody const*, statvfs* buf); ErrorOr> find_elf_interpreter_for_executable(StringView path, Elf_Ehdr const& main_executable_header, size_t main_executable_header_size, size_t file_size, Optional& minimum_stack_size); ErrorOr do_kill(Process&, int signal); ErrorOr do_killpg(ProcessGroupID pgrp, int signal); ErrorOr do_killall(int signal); ErrorOr do_killself(int signal); ErrorOr do_waitid(Variant, NonnullRefPtr> waitee, int options); static ErrorOr> get_syscall_path_argument(Userspace user_path, size_t path_length); static ErrorOr> get_syscall_path_argument(Syscall::StringArgument const&); bool has_tracee_thread(ProcessID tracer_pid); void clear_signal_handlers_for_exec(); void clear_futex_queues_on_exec(); ErrorOr get_futex_key(FlatPtr user_address, bool shared); ErrorOr remap_range_as_stack(FlatPtr address, size_t size); ErrorOr open_impl(Userspace); ErrorOr close_impl(int fd); ErrorOr read_impl(int fd, Userspace buffer, size_t size); ErrorOr pread_impl(int fd, Userspace, size_t, off_t); ErrorOr readv_impl(int fd, Userspace iov, int iov_count); public: ErrorOr traverse_as_directory(FileSystemID, Function(FileSystem::DirectoryEntryView const&)> callback) const; ErrorOr> lookup_as_directory(ProcFS&, StringView name) const; ErrorOr procfs_get_fds_stats(KBufferBuilder& builder) const; ErrorOr procfs_get_perf_events(KBufferBuilder& builder) const; ErrorOr procfs_get_unveil_stats(KBufferBuilder& builder) const; ErrorOr procfs_get_pledge_stats(KBufferBuilder& builder) const; ErrorOr procfs_get_virtual_memory_stats(KBufferBuilder& builder) const; ErrorOr procfs_get_binary_link(KBufferBuilder& builder) const; ErrorOr procfs_get_current_work_directory_link(KBufferBuilder& builder) const; ErrorOr procfs_get_command_line(KBufferBuilder& builder) const; mode_t binary_link_required_mode() const; ErrorOr procfs_get_thread_stack(ThreadID thread_id, KBufferBuilder& builder) const; ErrorOr traverse_stacks_directory(FileSystemID, Function(FileSystem::DirectoryEntryView const&)> callback) const; ErrorOr> lookup_stacks_directory(ProcFS&, StringView name) const; ErrorOr procfs_get_file_description_link(unsigned fd, KBufferBuilder& builder) const; ErrorOr traverse_file_descriptions_directory(FileSystemID, Function(FileSystem::DirectoryEntryView const&)> callback) const; ErrorOr> lookup_file_descriptions_directory(ProcFS&, StringView name) const; ErrorOr> lookup_children_directory(ProcFS&, StringView name) const; ErrorOr traverse_children_directory(FileSystemID, Function(FileSystem::DirectoryEntryView const&)> callback) const; ErrorOr procfs_get_child_process_link(ProcessID child_pid, KBufferBuilder& builder) const; private: inline PerformanceEventBuffer* current_perf_events_buffer() { if (g_profiling_all_threads) return g_global_perf_events; if (m_profiling) return m_perf_event_buffer.ptr(); return nullptr; } SpinlockProtected m_name; SpinlockProtected, LockRank::None> m_space; RecursiveSpinlock mutable m_protected_data_lock; AtomicEdgeAction m_protected_data_refs; void protect_data(); void unprotect_data(); OwnPtr m_tracer; public: class OpenFileDescriptionAndFlags { public: bool is_valid() const { return !m_description.is_null(); } bool is_allocated() const { return m_is_allocated; } void allocate() { VERIFY(!m_is_allocated); VERIFY(!is_valid()); m_is_allocated = true; } void deallocate() { VERIFY(m_is_allocated); VERIFY(!is_valid()); m_is_allocated = false; } OpenFileDescription* description() { return m_description; } OpenFileDescription const* description() const { return m_description; } u32 flags() const { return m_flags; } void set_flags(u32 flags) { m_flags = flags; } void clear(); void set(NonnullRefPtr, u32 flags = 0); private: RefPtr m_description; bool m_is_allocated { false }; u32 m_flags { 0 }; }; class ScopedDescriptionAllocation; class OpenFileDescriptions { AK_MAKE_NONCOPYABLE(OpenFileDescriptions); AK_MAKE_NONMOVABLE(OpenFileDescriptions); friend class Process; public: OpenFileDescriptions() { } ALWAYS_INLINE OpenFileDescriptionAndFlags const& operator[](size_t i) const { return at(i); } ALWAYS_INLINE OpenFileDescriptionAndFlags& operator[](size_t i) { return at(i); } ErrorOr try_clone(Kernel::Process::OpenFileDescriptions const& other) { TRY(try_resize(other.m_fds_metadatas.size())); for (size_t i = 0; i < other.m_fds_metadatas.size(); ++i) { m_fds_metadatas[i] = other.m_fds_metadatas[i]; } return {}; } OpenFileDescriptionAndFlags const& at(size_t i) const; OpenFileDescriptionAndFlags& at(size_t i); OpenFileDescriptionAndFlags const* get_if_valid(size_t i) const; OpenFileDescriptionAndFlags* get_if_valid(size_t i); void enumerate(Function) const; ErrorOr try_enumerate(Function(OpenFileDescriptionAndFlags const&)>) const; void change_each(Function); ErrorOr allocate(int first_candidate_fd = 0); size_t open_count() const; ErrorOr try_resize(size_t size) { return m_fds_metadatas.try_resize(size); } static constexpr size_t max_open() { return s_max_open_file_descriptors; } void clear() { m_fds_metadatas.clear(); } ErrorOr> open_file_description(int fd) const; private: static constexpr size_t s_max_open_file_descriptors { FD_SETSIZE }; Vector m_fds_metadatas; }; class ScopedDescriptionAllocation { AK_MAKE_NONCOPYABLE(ScopedDescriptionAllocation); public: ScopedDescriptionAllocation() = default; ScopedDescriptionAllocation(int tracked_fd, OpenFileDescriptionAndFlags* description) : fd(tracked_fd) , m_description(description) { } ScopedDescriptionAllocation(ScopedDescriptionAllocation&& other) : fd(other.fd) { // Take over the responsibility of tracking to deallocation. swap(m_description, other.m_description); } ScopedDescriptionAllocation& operator=(ScopedDescriptionAllocation&& other) { if (this != &other) { m_description = exchange(other.m_description, nullptr); fd = exchange(other.fd, -1); } return *this; } ~ScopedDescriptionAllocation() { if (m_description && m_description->is_allocated() && !m_description->is_valid()) { m_description->deallocate(); } } int fd { -1 }; private: OpenFileDescriptionAndFlags* m_description { nullptr }; }; MutexProtected& fds() { return m_fds; } MutexProtected const& fds() const { return m_fds; } ErrorOr> open_file_description(int fd) { return m_fds.with_shared([fd](auto& fds) { return fds.open_file_description(fd); }); } ErrorOr> open_file_description_ignoring_negative(int fd) { if (fd < 0) return nullptr; return open_file_description(fd); } ErrorOr> open_file_description(int fd) const { return m_fds.with_shared([fd](auto& fds) { return fds.open_file_description(fd); }); } ErrorOr> open_file_description_ignoring_negative(int fd) const { if (fd < 0) return nullptr; return open_file_description(fd); } ErrorOr allocate_fd() { return m_fds.with_exclusive([](auto& fds) { return fds.allocate(); }); } ErrorOr> custody_for_dirfd(Badge, int dirfd); private: ErrorOr> custody_for_dirfd(int dirfd); SpinlockProtected& thread_list() { return m_thread_list; } SpinlockProtected const& thread_list() const { return m_thread_list; } ErrorOr> get_thread_from_pid_or_tid(pid_t pid_or_tid, Syscall::SchedulerParametersMode mode); ErrorOr> get_thread_from_thread_list(pid_t tid); SpinlockProtected m_thread_list {}; MutexProtected m_fds; bool const m_is_kernel_process; Atomic m_state { State::Running }; bool m_profiling { false }; Atomic m_is_stopped { false }; bool m_should_generate_coredump { false }; #ifdef ENABLE_KERNEL_COVERAGE_COLLECTION KCOVInstance* m_kcov_instance { nullptr }; #endif SpinlockProtected, LockRank::None> m_executable; SpinlockProtected, LockRank::None> m_current_directory; UnixDateTime const m_creation_time; Vector> m_arguments; Vector> m_environment; IntrusiveListNode m_jail_process_list_node; IntrusiveListNode m_all_processes_list_node; public: using AllProcessesList = IntrusiveListRelaxedConst<&Process::m_all_processes_list_node>; using JailProcessList = IntrusiveListRelaxedConst<&Process::m_jail_process_list_node>; private: SpinlockProtected, LockRank::None> m_jail_process_list; SpinlockProtected, LockRank::Process> m_attached_jail {}; Mutex m_big_lock { "Process"sv, Mutex::MutexBehavior::BigLock }; Mutex m_ptrace_lock { "ptrace"sv }; SpinlockProtected, LockRank::None> m_alarm_timer; SpinlockProtected m_unveil_data; SpinlockProtected m_exec_unveil_data; OwnPtr m_perf_event_buffer; // This member is used in the implementation of ptrace's PT_TRACEME flag. // If it is set to true, the process will stop at the next execve syscall // and wait for a tracer to attach. bool m_wait_for_tracer_at_next_execve { false }; Thread::WaitBlockerSet m_wait_blocker_set; struct CoredumpProperty { OwnPtr key; OwnPtr value; }; SpinlockProtected, LockRank::None> m_coredump_properties {}; Vector> m_threads_for_coredump; struct SignalActionData { VirtualAddress handler_or_sigaction; int flags { 0 }; u32 mask { 0 }; }; Array m_signal_action_data; static_assert(sizeof(ProtectedValues) < (PAGE_SIZE)); alignas(4096) ProtectedValues m_protected_values_do_not_access_directly; u8 m_protected_values_padding[PAGE_SIZE - sizeof(ProtectedValues)]; public: static SpinlockProtected& all_instances(); }; class ProcessList : public RefCounted { public: static ErrorOr> create(); SpinlockProtected& attached_processes() { return m_attached_processes; } SpinlockProtected const& attached_processes() const { return m_attached_processes; } private: ProcessList() = default; SpinlockProtected m_attached_processes; }; // Note: Process object should be 2 pages of 4096 bytes each. // It's not expected that the Process object will expand further because the first // page is used for all unprotected values (which should be plenty of space for them). // The second page is being used exclusively for write-protected values. static_assert(AssertSize()); extern RecursiveSpinlock g_profiling_lock; template Callback> inline IterationDecision Process::for_each_thread(Callback callback) { return thread_list().with([&](auto& thread_list) -> IterationDecision { for (auto& thread : thread_list) { IterationDecision decision = callback(thread); if (decision != IterationDecision::Continue) return decision; } return IterationDecision::Continue; }); } template Callback> inline void Process::for_each_ignoring_jails(Callback callback) { Process::all_instances().with([&](auto const& list) { for (auto it = list.begin(); it != list.end();) { auto& process = *it; ++it; if (callback(process) == IterationDecision::Break) break; } }); } template Callback> inline IterationDecision Process::for_each_thread(Callback callback) const { return thread_list().with([&](auto& thread_list) -> IterationDecision { for (auto& thread : thread_list) { IterationDecision decision = callback(thread); if (decision != IterationDecision::Continue) return decision; } return IterationDecision::Continue; }); } template Callback> inline IterationDecision Process::for_each_thread(Callback callback) const { thread_list().with([&](auto& thread_list) { for (auto& thread : thread_list) callback(thread); }); return IterationDecision::Continue; } inline ErrorOr Process::try_for_each_thread(Function(Thread const&)> callback) const { return thread_list().with([&](auto& thread_list) -> ErrorOr { for (auto& thread : thread_list) TRY(callback(thread)); return {}; }); } template Callback> inline IterationDecision Process::for_each_thread(Callback callback) { thread_list().with([&](auto& thread_list) { for (auto& thread : thread_list) callback(thread); }); return IterationDecision::Continue; } inline ProcessID Thread::pid() const { return m_process->pid(); } } #define VERIFY_PROCESS_BIG_LOCK_ACQUIRED(process) \ VERIFY(process->big_lock().is_exclusively_locked_by_current_thread()) #define VERIFY_NO_PROCESS_BIG_LOCK(process) \ VERIFY(!process->big_lock().is_exclusively_locked_by_current_thread()) inline ErrorOr> try_copy_kstring_from_user(Kernel::Syscall::StringArgument const& string) { Userspace characters((FlatPtr)string.characters); return try_copy_kstring_from_user(characters, string.length); } template<> struct AK::Formatter : AK::Formatter { ErrorOr format(FormatBuilder& builder, Kernel::Process const& value) { return value.name().with([&](auto& process_name) { return AK::Formatter::format(builder, "{}({})"sv, process_name.representable_view(), value.pid().value()); }); } }; namespace AK { template<> struct Traits : public DefaultTraits { static unsigned hash(Kernel::GlobalFutexKey const& futex_key) { return pair_int_hash(ptr_hash(futex_key.raw.parent), ptr_hash(futex_key.raw.offset)); } static bool equals(Kernel::GlobalFutexKey const& a, Kernel::GlobalFutexKey const& b) { return a.raw.parent == b.raw.parent && a.raw.offset == b.raw.offset; } }; };