/* SPDX-License-Identifier: LGPL-2.1-or-later */ #include #include #include "alloc-util.h" #include "bus-common-errors.h" #include "bus-error.h" #include "bus-locator.h" #include "bus-util.h" #include "cgroup-util.h" #include "clean-ipc.h" #include "env-file.h" #include "escape.h" #include "fd-util.h" #include "fileio.h" #include "format-util.h" #include "fs-util.h" #include "hashmap.h" #include "label-util.h" #include "limits-util.h" #include "logind-dbus.h" #include "logind-user-dbus.h" #include "logind-user.h" #include "mkdir-label.h" #include "parse-util.h" #include "path-util.h" #include "percent-util.h" #include "rm-rf.h" #include "serialize.h" #include "special.h" #include "stdio-util.h" #include "string-table.h" #include "strv.h" #include "tmpfile-util.h" #include "uid-alloc-range.h" #include "unit-name.h" #include "user-util.h" int user_new(User **ret, Manager *m, UserRecord *ur) { _cleanup_(user_freep) User *u = NULL; char lu[DECIMAL_STR_MAX(uid_t) + 1]; int r; assert(ret); assert(m); assert(ur); if (!ur->user_name) return -EINVAL; if (!uid_is_valid(ur->uid)) return -EINVAL; u = new(User, 1); if (!u) return -ENOMEM; *u = (User) { .manager = m, .user_record = user_record_ref(ur), .last_session_timestamp = USEC_INFINITY, }; if (asprintf(&u->state_file, "/run/systemd/users/" UID_FMT, ur->uid) < 0) return -ENOMEM; if (asprintf(&u->runtime_path, "/run/user/" UID_FMT, ur->uid) < 0) return -ENOMEM; xsprintf(lu, UID_FMT, ur->uid); r = slice_build_subslice(SPECIAL_USER_SLICE, lu, &u->slice); if (r < 0) return r; r = unit_name_build("user", lu, ".service", &u->service); if (r < 0) return r; r = unit_name_build("user-runtime-dir", lu, ".service", &u->runtime_dir_service); if (r < 0) return r; r = hashmap_put(m->users, UID_TO_PTR(ur->uid), u); if (r < 0) return r; r = hashmap_put(m->user_units, u->slice, u); if (r < 0) return r; r = hashmap_put(m->user_units, u->service, u); if (r < 0) return r; r = hashmap_put(m->user_units, u->runtime_dir_service, u); if (r < 0) return r; *ret = TAKE_PTR(u); return 0; } User *user_free(User *u) { if (!u) return NULL; if (u->in_gc_queue) LIST_REMOVE(gc_queue, u->manager->user_gc_queue, u); while (u->sessions) session_free(u->sessions); if (u->service) hashmap_remove_value(u->manager->user_units, u->service, u); if (u->runtime_dir_service) hashmap_remove_value(u->manager->user_units, u->runtime_dir_service, u); if (u->slice) hashmap_remove_value(u->manager->user_units, u->slice, u); hashmap_remove_value(u->manager->users, UID_TO_PTR(u->user_record->uid), u); sd_event_source_unref(u->timer_event_source); u->service_job = mfree(u->service_job); u->service = mfree(u->service); u->runtime_dir_service = mfree(u->runtime_dir_service); u->slice = mfree(u->slice); u->runtime_path = mfree(u->runtime_path); u->state_file = mfree(u->state_file); user_record_unref(u->user_record); return mfree(u); } static int user_save_internal(User *u) { _cleanup_(unlink_and_freep) char *temp_path = NULL; _cleanup_fclose_ FILE *f = NULL; int r; assert(u); assert(u->state_file); r = mkdir_safe_label("/run/systemd/users", 0755, 0, 0, MKDIR_WARN_MODE); if (r < 0) goto fail; r = fopen_temporary(u->state_file, &f, &temp_path); if (r < 0) goto fail; (void) fchmod(fileno(f), 0644); fprintf(f, "# This is private data. Do not parse.\n" "NAME=%s\n" "STATE=%s\n" /* friendly user-facing state */ "STOPPING=%s\n", /* low-level state */ u->user_record->user_name, user_state_to_string(user_get_state(u)), yes_no(u->stopping)); /* LEGACY: no-one reads RUNTIME= anymore, drop it at some point */ if (u->runtime_path) fprintf(f, "RUNTIME=%s\n", u->runtime_path); if (u->service_job) fprintf(f, "SERVICE_JOB=%s\n", u->service_job); if (u->display) fprintf(f, "DISPLAY=%s\n", u->display->id); if (dual_timestamp_is_set(&u->timestamp)) fprintf(f, "REALTIME="USEC_FMT"\n" "MONOTONIC="USEC_FMT"\n", u->timestamp.realtime, u->timestamp.monotonic); if (u->last_session_timestamp != USEC_INFINITY) fprintf(f, "LAST_SESSION_TIMESTAMP=" USEC_FMT "\n", u->last_session_timestamp); if (u->sessions) { bool first; fputs("SESSIONS=", f); first = true; LIST_FOREACH(sessions_by_user, i, u->sessions) { if (first) first = false; else fputc(' ', f); fputs(i->id, f); } fputs("\nSEATS=", f); first = true; LIST_FOREACH(sessions_by_user, i, u->sessions) { if (!i->seat) continue; if (first) first = false; else fputc(' ', f); fputs(i->seat->id, f); } fputs("\nACTIVE_SESSIONS=", f); first = true; LIST_FOREACH(sessions_by_user, i, u->sessions) { if (!session_is_active(i)) continue; if (first) first = false; else fputc(' ', f); fputs(i->id, f); } fputs("\nONLINE_SESSIONS=", f); first = true; LIST_FOREACH(sessions_by_user, i, u->sessions) { if (session_get_state(i) == SESSION_CLOSING) continue; if (first) first = false; else fputc(' ', f); fputs(i->id, f); } fputs("\nACTIVE_SEATS=", f); first = true; LIST_FOREACH(sessions_by_user, i, u->sessions) { if (!session_is_active(i) || !i->seat) continue; if (first) first = false; else fputc(' ', f); fputs(i->seat->id, f); } fputs("\nONLINE_SEATS=", f); first = true; LIST_FOREACH(sessions_by_user, i, u->sessions) { if (session_get_state(i) == SESSION_CLOSING || !i->seat) continue; if (first) first = false; else fputc(' ', f); fputs(i->seat->id, f); } fputc('\n', f); } r = fflush_and_check(f); if (r < 0) goto fail; if (rename(temp_path, u->state_file) < 0) { r = -errno; goto fail; } temp_path = mfree(temp_path); return 0; fail: (void) unlink(u->state_file); return log_error_errno(r, "Failed to save user data %s: %m", u->state_file); } int user_save(User *u) { assert(u); if (!u->started) return 0; return user_save_internal(u); } int user_load(User *u) { _cleanup_free_ char *realtime = NULL, *monotonic = NULL, *stopping = NULL, *last_session_timestamp = NULL; int r; assert(u); r = parse_env_file(NULL, u->state_file, "SERVICE_JOB", &u->service_job, "STOPPING", &stopping, "REALTIME", &realtime, "MONOTONIC", &monotonic, "LAST_SESSION_TIMESTAMP", &last_session_timestamp); if (r == -ENOENT) return 0; if (r < 0) return log_error_errno(r, "Failed to read %s: %m", u->state_file); if (stopping) { r = parse_boolean(stopping); if (r < 0) log_debug_errno(r, "Failed to parse 'STOPPING' boolean: %s", stopping); else u->stopping = r; } if (realtime) (void) deserialize_usec(realtime, &u->timestamp.realtime); if (monotonic) (void) deserialize_usec(monotonic, &u->timestamp.monotonic); if (last_session_timestamp) (void) deserialize_usec(last_session_timestamp, &u->last_session_timestamp); return 0; } static void user_start_service(User *u) { _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; int r; assert(u); /* Start the service containing the "systemd --user" instance (user@.service). Note that we don't explicitly * start the per-user slice or the systemd-runtime-dir@.service instance, as those are pulled in both by * user@.service and the session scopes as dependencies. */ u->service_job = mfree(u->service_job); r = manager_start_unit(u->manager, u->service, &error, &u->service_job); if (r < 0) log_full_errno(sd_bus_error_has_name(&error, BUS_ERROR_UNIT_MASKED) ? LOG_DEBUG : LOG_WARNING, r, "Failed to start user service '%s', ignoring: %s", u->service, bus_error_message(&error, r)); } static int update_slice_callback(sd_bus_message *m, void *userdata, sd_bus_error *ret_error) { _cleanup_(user_record_unrefp) UserRecord *ur = ASSERT_PTR(userdata); const sd_bus_error *e; int r; assert(m); e = sd_bus_message_get_error(m); if (e) { r = sd_bus_error_get_errno(e); log_warning_errno(r, "Failed to update slice of %s, ignoring: %s", ur->user_name, bus_error_message(e, r)); return 0; } log_debug("Successfully set slice parameters of %s.", ur->user_name); return 0; } static int user_update_slice(User *u) { _cleanup_(sd_bus_message_unrefp) sd_bus_message *m = NULL; int r; assert(u); if (u->user_record->tasks_max == UINT64_MAX && u->user_record->memory_high == UINT64_MAX && u->user_record->memory_max == UINT64_MAX && u->user_record->cpu_weight == UINT64_MAX && u->user_record->io_weight == UINT64_MAX) return 0; r = bus_message_new_method_call(u->manager->bus, &m, bus_systemd_mgr, "SetUnitProperties"); if (r < 0) return bus_log_create_error(r); r = sd_bus_message_append(m, "sb", u->slice, true); if (r < 0) return bus_log_create_error(r); r = sd_bus_message_open_container(m, 'a', "(sv)"); if (r < 0) return bus_log_create_error(r); const struct { const char *name; uint64_t value; } settings[] = { { "TasksMax", u->user_record->tasks_max }, { "MemoryMax", u->user_record->memory_max }, { "MemoryHigh", u->user_record->memory_high }, { "CPUWeight", u->user_record->cpu_weight }, { "IOWeight", u->user_record->io_weight }, }; for (size_t i = 0; i < ELEMENTSOF(settings); i++) if (settings[i].value != UINT64_MAX) { r = sd_bus_message_append(m, "(sv)", settings[i].name, "t", settings[i].value); if (r < 0) return bus_log_create_error(r); } r = sd_bus_message_close_container(m); if (r < 0) return bus_log_create_error(r); r = sd_bus_call_async(u->manager->bus, NULL, m, update_slice_callback, u->user_record, 0); if (r < 0) return log_error_errno(r, "Failed to change user slice properties: %m"); /* Ref the user record pointer, so that the slot keeps it pinned */ user_record_ref(u->user_record); return 0; } int user_start(User *u) { assert(u); if (u->started && !u->stopping) return 0; /* If u->stopping is set, the user is marked for removal and service stop-jobs are queued. We have to clear * that flag before queueing the start-jobs again. If they succeed, the user object can be re-used just fine * (pid1 takes care of job-ordering and proper restart), but if they fail, we want to force another user_stop() * so possibly pending units are stopped. */ u->stopping = false; if (!u->started) log_debug("Starting services for new user %s.", u->user_record->user_name); /* Save the user data so far, because pam_systemd will read the XDG_RUNTIME_DIR out of it while starting up * systemd --user. We need to do user_save_internal() because we have not "officially" started yet. */ user_save_internal(u); /* Set slice parameters */ (void) user_update_slice(u); /* Start user@UID.service */ user_start_service(u); if (!u->started) { if (!dual_timestamp_is_set(&u->timestamp)) dual_timestamp_get(&u->timestamp); user_send_signal(u, true); u->started = true; } /* Save new user data */ user_save(u); return 0; } static void user_stop_service(User *u, bool force) { _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; int r; assert(u); assert(u->service); /* The reverse of user_start_service(). Note that we only stop user@UID.service here, and let StopWhenUnneeded= * deal with the slice and the user-runtime-dir@.service instance. */ u->service_job = mfree(u->service_job); r = manager_stop_unit(u->manager, u->service, force ? "replace" : "fail", &error, &u->service_job); if (r < 0) log_warning_errno(r, "Failed to stop user service '%s', ignoring: %s", u->service, bus_error_message(&error, r)); } int user_stop(User *u, bool force) { int r = 0; assert(u); /* This is called whenever we begin with tearing down a user record. It's called in two cases: explicit API * request to do so via the bus (in which case 'force' is true) and automatically due to GC, if there's no * session left pinning it (in which case 'force' is false). Note that this just initiates tearing down of the * user, the User object will remain in memory until user_finalize() is called, see below. */ if (!u->started) return 0; if (u->stopping) { /* Stop jobs have already been queued */ user_save(u); return 0; } LIST_FOREACH(sessions_by_user, s, u->sessions) { int k; k = session_stop(s, force); if (k < 0) r = k; } user_stop_service(u, force); u->stopping = true; user_save(u); return r; } int user_finalize(User *u) { int r = 0, k; assert(u); /* Called when the user is really ready to be freed, i.e. when all unit stop jobs and suchlike for it are * done. This is called as a result of an earlier user_done() when all jobs are completed. */ if (u->started) log_debug("User %s logged out.", u->user_record->user_name); LIST_FOREACH(sessions_by_user, s, u->sessions) { k = session_finalize(s); if (k < 0) r = k; } /* Clean SysV + POSIX IPC objects, but only if this is not a system user. Background: in many setups cronjobs * are run in full PAM and thus logind sessions, even if the code run doesn't belong to actual users but to * system components. Since enable RemoveIPC= globally for all users, we need to be a bit careful with such * cases, as we shouldn't accidentally remove a system service's IPC objects while it is running, just because * a cronjob running as the same user just finished. Hence: exclude system users generally from IPC clean-up, * and do it only for normal users. */ if (u->manager->remove_ipc && !uid_is_system(u->user_record->uid)) { k = clean_ipc_by_uid(u->user_record->uid); if (k < 0) r = k; } (void) unlink(u->state_file); user_add_to_gc_queue(u); if (u->started) { user_send_signal(u, false); u->started = false; } return r; } int user_get_idle_hint(User *u, dual_timestamp *t) { bool idle_hint = true; dual_timestamp ts = DUAL_TIMESTAMP_NULL; assert(u); LIST_FOREACH(sessions_by_user, s, u->sessions) { dual_timestamp k; int ih; ih = session_get_idle_hint(s, &k); if (ih < 0) return ih; if (!ih) { if (!idle_hint) { if (k.monotonic < ts.monotonic) ts = k; } else { idle_hint = false; ts = k; } } else if (idle_hint) { if (k.monotonic > ts.monotonic) ts = k; } } if (t) *t = ts; return idle_hint; } int user_check_linger_file(User *u) { _cleanup_free_ char *cc = NULL; char *p = NULL; cc = cescape(u->user_record->user_name); if (!cc) return -ENOMEM; p = strjoina("/var/lib/systemd/linger/", cc); if (access(p, F_OK) < 0) { if (errno != ENOENT) return -errno; return false; } return true; } static bool user_unit_active(User *u) { int r; assert(u->service); assert(u->runtime_dir_service); assert(u->slice); FOREACH_STRING(i, u->service, u->runtime_dir_service, u->slice) { _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; r = manager_unit_is_active(u->manager, i, &error); if (r < 0) log_debug_errno(r, "Failed to determine whether unit '%s' is active, ignoring: %s", i, bus_error_message(&error, r)); if (r != 0) return true; } return false; } static usec_t user_get_stop_delay(User *u) { assert(u); if (u->user_record->stop_delay_usec != UINT64_MAX) return u->user_record->stop_delay_usec; if (user_record_removable(u->user_record) > 0) return 0; /* For removable users lower the stop delay to zero */ return u->manager->user_stop_delay; } bool user_may_gc(User *u, bool drop_not_started) { int r; assert(u); if (drop_not_started && !u->started) return true; if (u->sessions) return false; if (u->last_session_timestamp != USEC_INFINITY) { usec_t user_stop_delay; /* All sessions have been closed. Let's see if we shall leave the user record around for a bit */ user_stop_delay = user_get_stop_delay(u); if (user_stop_delay == USEC_INFINITY) return false; /* Leave it around forever! */ if (user_stop_delay > 0 && now(CLOCK_MONOTONIC) < usec_add(u->last_session_timestamp, user_stop_delay)) return false; /* Leave it around for a bit longer. */ } /* Is this a user that shall stay around forever ("linger")? Before we say "no" to GC'ing for lingering users, let's check * if any of the three units that we maintain for this user is still around. If none of them is, * there's no need to keep this user around even if lingering is enabled. */ if (user_check_linger_file(u) > 0 && user_unit_active(u)) return false; /* Check if our job is still pending */ if (u->service_job) { _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL; r = manager_job_is_active(u->manager, u->service_job, &error); if (r < 0) log_debug_errno(r, "Failed to determine whether job '%s' is pending, ignoring: %s", u->service_job, bus_error_message(&error, r)); if (r != 0) return false; } /* Note that we don't care if the three units we manage for each user object are up or not, as we are managing * their state rather than tracking it. */ return true; } void user_add_to_gc_queue(User *u) { assert(u); if (u->in_gc_queue) return; LIST_PREPEND(gc_queue, u->manager->user_gc_queue, u); u->in_gc_queue = true; } UserState user_get_state(User *u) { assert(u); if (u->stopping) return USER_CLOSING; if (!u->started || u->service_job) return USER_OPENING; if (u->sessions) { bool all_closing = true; LIST_FOREACH(sessions_by_user, i, u->sessions) { SessionState state; state = session_get_state(i); if (state == SESSION_ACTIVE) return USER_ACTIVE; if (state != SESSION_CLOSING) all_closing = false; } return all_closing ? USER_CLOSING : USER_ONLINE; } if (user_check_linger_file(u) > 0 && user_unit_active(u)) return USER_LINGERING; return USER_CLOSING; } int user_kill(User *u, int signo) { assert(u); return manager_kill_unit(u->manager, u->slice, KILL_ALL, signo, NULL); } static bool elect_display_filter(Session *s) { /* Return true if the session is a candidate for the user’s ‘primary session’ or ‘display’. */ assert(s); return IN_SET(s->class, SESSION_USER, SESSION_GREETER) && s->started && !s->stopping; } static int elect_display_compare(Session *s1, Session *s2) { /* Indexed by SessionType. Lower numbers mean more preferred. */ static const int type_ranks[_SESSION_TYPE_MAX] = { [SESSION_UNSPECIFIED] = 0, [SESSION_TTY] = -2, [SESSION_X11] = -3, [SESSION_WAYLAND] = -3, [SESSION_MIR] = -3, [SESSION_WEB] = -1, }; /* Calculate the partial order relationship between s1 and s2, * returning < 0 if s1 is preferred as the user’s ‘primary session’, * 0 if s1 and s2 are equally preferred or incomparable, or > 0 if s2 * is preferred. * * s1 or s2 may be NULL. */ if (!s1 && !s2) return 0; if ((s1 == NULL) != (s2 == NULL)) return (s1 == NULL) - (s2 == NULL); if (s1->stopping != s2->stopping) return s1->stopping - s2->stopping; if ((s1->class != SESSION_USER) != (s2->class != SESSION_USER)) return (s1->class != SESSION_USER) - (s2->class != SESSION_USER); if ((s1->type == _SESSION_TYPE_INVALID) != (s2->type == _SESSION_TYPE_INVALID)) return (s1->type == _SESSION_TYPE_INVALID) - (s2->type == _SESSION_TYPE_INVALID); if (s1->type != s2->type) return type_ranks[s1->type] - type_ranks[s2->type]; return 0; } void user_elect_display(User *u) { assert(u); /* This elects a primary session for each user, which we call the "display". We try to keep the assignment * stable, but we "upgrade" to better choices. */ log_debug("Electing new display for user %s", u->user_record->user_name); LIST_FOREACH(sessions_by_user, s, u->sessions) { if (!elect_display_filter(s)) { log_debug("Ignoring session %s", s->id); continue; } if (elect_display_compare(s, u->display) < 0) { log_debug("Choosing session %s in preference to %s", s->id, u->display ? u->display->id : "-"); u->display = s; } } } static int user_stop_timeout_callback(sd_event_source *es, uint64_t usec, void *userdata) { User *u = ASSERT_PTR(userdata); user_add_to_gc_queue(u); return 0; } void user_update_last_session_timer(User *u) { usec_t user_stop_delay; int r; assert(u); if (u->sessions) { /* There are sessions, turn off the timer */ u->last_session_timestamp = USEC_INFINITY; u->timer_event_source = sd_event_source_unref(u->timer_event_source); return; } if (u->last_session_timestamp != USEC_INFINITY) return; /* Timer already started */ u->last_session_timestamp = now(CLOCK_MONOTONIC); assert(!u->timer_event_source); user_stop_delay = user_get_stop_delay(u); if (!timestamp_is_set(user_stop_delay)) return; if (sd_event_get_state(u->manager->event) == SD_EVENT_FINISHED) { log_debug("Not allocating user stop timeout, since we are already exiting."); return; } r = sd_event_add_time(u->manager->event, &u->timer_event_source, CLOCK_MONOTONIC, usec_add(u->last_session_timestamp, user_stop_delay), 0, user_stop_timeout_callback, u); if (r < 0) log_warning_errno(r, "Failed to enqueue user stop event source, ignoring: %m"); if (DEBUG_LOGGING) log_debug("Last session of user '%s' logged out, terminating user context in %s.", u->user_record->user_name, FORMAT_TIMESPAN(user_stop_delay, USEC_PER_MSEC)); } static const char* const user_state_table[_USER_STATE_MAX] = { [USER_OFFLINE] = "offline", [USER_OPENING] = "opening", [USER_LINGERING] = "lingering", [USER_ONLINE] = "online", [USER_ACTIVE] = "active", [USER_CLOSING] = "closing" }; DEFINE_STRING_TABLE_LOOKUP(user_state, UserState); int config_parse_tmpfs_size( const char* unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata) { uint64_t *sz = ASSERT_PTR(data); int r; assert(filename); assert(lvalue); assert(rvalue); /* First, try to parse as percentage */ r = parse_permyriad(rvalue); if (r > 0) *sz = physical_memory_scale(r, 10000U); else { uint64_t k; /* If the passed argument was not a percentage, or out of range, parse as byte size */ r = parse_size(rvalue, 1024, &k); if (r >= 0 && (k <= 0 || (uint64_t) (size_t) k != k)) r = -ERANGE; if (r < 0) { log_syntax(unit, LOG_WARNING, filename, line, r, "Failed to parse size value '%s', ignoring: %m", rvalue); return 0; } *sz = PAGE_ALIGN((size_t) k); } return 0; } int config_parse_compat_user_tasks_max( const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata) { assert(filename); assert(lvalue); assert(rvalue); log_syntax(unit, LOG_NOTICE, filename, line, 0, "Support for option %s= has been removed.", lvalue); log_info("Hint: try creating /etc/systemd/system/user-.slice.d/50-limits.conf with:\n" " [Slice]\n" " TasksMax=%s", rvalue); return 0; }