1 /*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
4 This file is part of systemd.
6 Copyright 2010 Lennart Poettering
8 systemd is free software; you can redistribute it and/or modify it
9 under the terms of the GNU Lesser General Public License as published by
10 the Free Software Foundation; either version 2.1 of the License, or
11 (at your option) any later version.
13 systemd is distributed in the hope that it will be useful, but
14 WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 Lesser General Public License for more details.
18 You should have received a copy of the GNU Lesser General Public License
19 along with systemd; If not, see <http://www.gnu.org/licenses/>.
28 #include <sys/mount.h>
29 #include <sys/prctl.h>
30 #include <sys/reboot.h>
36 #ifdef HAVE_VALGRIND_VALGRIND_H
37 #include <valgrind/valgrind.h>
41 #include "sd-daemon.h"
43 #include "alloc-util.h"
44 #include "architecture.h"
46 #include "bus-error.h"
48 #include "capability-util.h"
49 #include "clock-util.h"
50 #include "conf-parser.h"
51 #include "cpu-set-util.h"
52 #include "dbus-manager.h"
58 #include "formats-util.h"
60 #include "hostname-setup.h"
61 #include "ima-setup.h"
63 #include "kmod-setup.h"
64 #include "load-fragment.h"
66 #include "loopback-setup.h"
67 #include "machine-id-setup.h"
70 #include "mount-setup.h"
72 #include "parse-util.h"
73 #include "proc-cmdline.h"
74 #include "process-util.h"
75 #include "rlimit-util.h"
76 #include "selinux-setup.h"
77 #include "selinux-util.h"
78 #include "signal-util.h"
79 #include "smack-setup.h"
81 #include "stat-util.h"
82 #include "stdio-util.h"
84 #include "switch-root.h"
85 #include "terminal-util.h"
86 #include "user-util.h"
95 ACTION_DUMP_CONFIGURATION_ITEMS
,
97 } arg_action
= ACTION_RUN
;
98 static char *arg_default_unit
= NULL
;
99 static ManagerRunningAs arg_running_as
= _MANAGER_RUNNING_AS_INVALID
;
100 static bool arg_dump_core
= true;
101 static int arg_crash_chvt
= -1;
102 static bool arg_crash_shell
= false;
103 static bool arg_crash_reboot
= false;
104 static bool arg_confirm_spawn
= false;
105 static ShowStatus arg_show_status
= _SHOW_STATUS_UNSET
;
106 static bool arg_switched_root
= false;
107 static int arg_no_pager
= -1;
108 static char ***arg_join_controllers
= NULL
;
109 static ExecOutput arg_default_std_output
= EXEC_OUTPUT_JOURNAL
;
110 static ExecOutput arg_default_std_error
= EXEC_OUTPUT_INHERIT
;
111 static usec_t arg_default_restart_usec
= DEFAULT_RESTART_USEC
;
112 static usec_t arg_default_timeout_start_usec
= DEFAULT_TIMEOUT_USEC
;
113 static usec_t arg_default_timeout_stop_usec
= DEFAULT_TIMEOUT_USEC
;
114 static usec_t arg_default_start_limit_interval
= DEFAULT_START_LIMIT_INTERVAL
;
115 static unsigned arg_default_start_limit_burst
= DEFAULT_START_LIMIT_BURST
;
116 static usec_t arg_runtime_watchdog
= 0;
117 static usec_t arg_shutdown_watchdog
= 10 * USEC_PER_MINUTE
;
118 static char **arg_default_environment
= NULL
;
119 static struct rlimit
*arg_default_rlimit
[_RLIMIT_MAX
] = {};
120 static uint64_t arg_capability_bounding_set
= CAP_ALL
;
121 static nsec_t arg_timer_slack_nsec
= NSEC_INFINITY
;
122 static usec_t arg_default_timer_accuracy_usec
= 1 * USEC_PER_MINUTE
;
123 static Set
* arg_syscall_archs
= NULL
;
124 static FILE* arg_serialization
= NULL
;
125 static bool arg_default_cpu_accounting
= false;
126 static bool arg_default_blockio_accounting
= false;
127 static bool arg_default_memory_accounting
= false;
128 static bool arg_default_tasks_accounting
= true;
129 static uint64_t arg_default_tasks_max
= UINT64_C(512);
130 static sd_id128_t arg_machine_id
= {};
132 static void pager_open_if_enabled(void) {
134 if (arg_no_pager
<= 0)
140 noreturn
static void freeze_or_reboot(void) {
142 if (arg_crash_reboot
) {
143 log_notice("Rebooting in 10s...");
146 log_notice("Rebooting now...");
147 (void) reboot(RB_AUTOBOOT
);
148 log_emergency_errno(errno
, "Failed to reboot: %m");
151 log_emergency("Freezing execution.");
155 noreturn
static void crash(int sig
) {
160 /* Pass this on immediately, if this is not PID 1 */
162 else if (!arg_dump_core
)
163 log_emergency("Caught <%s>, not dumping core.", signal_to_string(sig
));
165 sa
= (struct sigaction
) {
166 .sa_handler
= nop_signal_handler
,
167 .sa_flags
= SA_NOCLDSTOP
|SA_RESTART
,
170 /* We want to wait for the core process, hence let's enable SIGCHLD */
171 (void) sigaction(SIGCHLD
, &sa
, NULL
);
173 pid
= raw_clone(SIGCHLD
, NULL
);
175 log_emergency_errno(errno
, "Caught <%s>, cannot fork for core dump: %m", signal_to_string(sig
));
178 .rlim_cur
= RLIM_INFINITY
,
179 .rlim_max
= RLIM_INFINITY
,
182 /* Enable default signal handler for core dump */
183 sa
= (struct sigaction
) {
184 .sa_handler
= SIG_DFL
,
186 (void) sigaction(sig
, &sa
, NULL
);
188 /* Don't limit the core dump size */
189 (void) setrlimit(RLIMIT_CORE
, &rl
);
191 /* Just to be sure... */
194 /* Raise the signal again */
196 (void) kill(pid
, sig
); /* raise() would kill the parent */
198 assert_not_reached("We shouldn't be here...");
204 /* Order things nicely. */
205 r
= wait_for_terminate(pid
, &status
);
207 log_emergency_errno(r
, "Caught <%s>, waitpid() failed: %m", signal_to_string(sig
));
208 else if (status
.si_code
!= CLD_DUMPED
)
209 log_emergency("Caught <%s>, core dump failed (child "PID_FMT
", code=%s, status=%i/%s).",
210 signal_to_string(sig
),
211 pid
, sigchld_code_to_string(status
.si_code
),
213 strna(status
.si_code
== CLD_EXITED
214 ? exit_status_to_string(status
.si_status
, EXIT_STATUS_FULL
)
215 : signal_to_string(status
.si_status
)));
217 log_emergency("Caught <%s>, dumped core as pid "PID_FMT
".", signal_to_string(sig
), pid
);
221 if (arg_crash_chvt
>= 0)
222 (void) chvt(arg_crash_chvt
);
224 sa
= (struct sigaction
) {
225 .sa_handler
= SIG_IGN
,
226 .sa_flags
= SA_NOCLDSTOP
|SA_NOCLDWAIT
|SA_RESTART
,
229 /* Let the kernel reap children for us */
230 (void) sigaction(SIGCHLD
, &sa
, NULL
);
232 if (arg_crash_shell
) {
233 log_notice("Executing crash shell in 10s...");
236 pid
= raw_clone(SIGCHLD
, NULL
);
238 log_emergency_errno(errno
, "Failed to fork off crash shell: %m");
241 (void) make_console_stdio();
242 (void) execle("/bin/sh", "/bin/sh", NULL
, environ
);
244 log_emergency_errno(errno
, "execle() failed: %m");
247 log_info("Spawned crash shell as PID "PID_FMT
".", pid
);
248 (void) wait_for_terminate(pid
, NULL
);
255 static void install_crash_handler(void) {
256 static const struct sigaction sa
= {
258 .sa_flags
= SA_NODEFER
, /* So that we can raise the signal again from the signal handler */
262 /* We ignore the return value here, since, we don't mind if we
263 * cannot set up a crash handler */
264 r
= sigaction_many(&sa
, SIGNALS_CRASH_HANDLER
, -1);
266 log_debug_errno(r
, "I had trouble setting up the crash handler, ignoring: %m");
269 static int console_setup(void) {
270 _cleanup_close_
int tty_fd
= -1;
273 tty_fd
= open_terminal("/dev/console", O_WRONLY
|O_NOCTTY
|O_CLOEXEC
);
275 return log_error_errno(tty_fd
, "Failed to open /dev/console: %m");
277 /* We don't want to force text mode. plymouth may be showing
278 * pictures already from initrd. */
279 r
= reset_terminal_fd(tty_fd
, false);
281 return log_error_errno(r
, "Failed to reset /dev/console: %m");
286 static int parse_crash_chvt(const char *value
) {
289 if (safe_atoi(value
, &arg_crash_chvt
) >= 0)
292 b
= parse_boolean(value
);
297 arg_crash_chvt
= 0; /* switch to where kmsg goes */
299 arg_crash_chvt
= -1; /* turn off switching */
304 static int set_machine_id(const char *m
) {
306 if (sd_id128_from_string(m
, &arg_machine_id
) < 0)
309 if (sd_id128_is_null(arg_machine_id
))
315 static int parse_proc_cmdline_item(const char *key
, const char *value
) {
321 if (streq(key
, "systemd.unit") && value
) {
324 return free_and_strdup(&arg_default_unit
, value
);
326 } else if (streq(key
, "rd.systemd.unit") && value
) {
329 return free_and_strdup(&arg_default_unit
, value
);
331 } else if (streq(key
, "systemd.dump_core") && value
) {
333 r
= parse_boolean(value
);
335 log_warning("Failed to parse dump core switch %s. Ignoring.", value
);
339 } else if (streq(key
, "systemd.crash_chvt") && value
) {
341 if (parse_crash_chvt(value
) < 0)
342 log_warning("Failed to parse crash chvt switch %s. Ignoring.", value
);
344 } else if (streq(key
, "systemd.crash_shell") && value
) {
346 r
= parse_boolean(value
);
348 log_warning("Failed to parse crash shell switch %s. Ignoring.", value
);
352 } else if (streq(key
, "systemd.crash_reboot") && value
) {
354 r
= parse_boolean(value
);
356 log_warning("Failed to parse crash reboot switch %s. Ignoring.", value
);
358 arg_crash_reboot
= r
;
360 } else if (streq(key
, "systemd.confirm_spawn") && value
) {
362 r
= parse_boolean(value
);
364 log_warning("Failed to parse confirm spawn switch %s. Ignoring.", value
);
366 arg_confirm_spawn
= r
;
368 } else if (streq(key
, "systemd.show_status") && value
) {
370 r
= parse_show_status(value
, &arg_show_status
);
372 log_warning("Failed to parse show status switch %s. Ignoring.", value
);
374 } else if (streq(key
, "systemd.default_standard_output") && value
) {
376 r
= exec_output_from_string(value
);
378 log_warning("Failed to parse default standard output switch %s. Ignoring.", value
);
380 arg_default_std_output
= r
;
382 } else if (streq(key
, "systemd.default_standard_error") && value
) {
384 r
= exec_output_from_string(value
);
386 log_warning("Failed to parse default standard error switch %s. Ignoring.", value
);
388 arg_default_std_error
= r
;
390 } else if (streq(key
, "systemd.setenv") && value
) {
392 if (env_assignment_is_valid(value
)) {
395 env
= strv_env_set(arg_default_environment
, value
);
397 arg_default_environment
= env
;
399 log_warning_errno(ENOMEM
, "Setting environment variable '%s' failed, ignoring: %m", value
);
401 log_warning("Environment variable name '%s' is not valid. Ignoring.", value
);
403 } else if (streq(key
, "systemd.machine_id") && value
) {
405 r
= set_machine_id(value
);
407 log_warning("MachineID '%s' is not valid. Ignoring.", value
);
409 } else if (streq(key
, "quiet") && !value
) {
411 if (arg_show_status
== _SHOW_STATUS_UNSET
)
412 arg_show_status
= SHOW_STATUS_AUTO
;
414 } else if (streq(key
, "debug") && !value
) {
416 /* Note that log_parse_environment() handles 'debug'
417 * too, and sets the log level to LOG_DEBUG. */
419 if (detect_container() > 0)
420 log_set_target(LOG_TARGET_CONSOLE
);
422 } else if (!in_initrd() && !value
) {
425 /* SysV compatibility */
426 target
= runlevel_to_target(key
);
428 return free_and_strdup(&arg_default_unit
, target
);
434 #define DEFINE_SETTER(name, func, descr) \
435 static int name(const char *unit, \
436 const char *filename, \
438 const char *section, \
439 unsigned section_line, \
440 const char *lvalue, \
442 const char *rvalue, \
454 log_syntax(unit, LOG_ERR, filename, line, r, \
455 "Invalid " descr "'%s': %m", \
461 DEFINE_SETTER(config_parse_level2
, log_set_max_level_from_string
, "log level")
462 DEFINE_SETTER(config_parse_target
, log_set_target_from_string
, "target")
463 DEFINE_SETTER(config_parse_color
, log_show_color_from_string
, "color" )
464 DEFINE_SETTER(config_parse_location
, log_show_location_from_string
, "location")
466 static int config_parse_cpu_affinity2(
468 const char *filename
,
471 unsigned section_line
,
478 _cleanup_cpu_free_ cpu_set_t
*c
= NULL
;
481 ncpus
= parse_cpu_set_and_warn(rvalue
, &c
, unit
, filename
, line
, lvalue
);
485 if (sched_setaffinity(0, CPU_ALLOC_SIZE(ncpus
), c
) < 0)
486 log_warning("Failed to set CPU affinity: %m");
491 static int config_parse_show_status(
493 const char *filename
,
496 unsigned section_line
,
504 ShowStatus
*b
= data
;
511 k
= parse_show_status(rvalue
, b
);
513 log_syntax(unit
, LOG_ERR
, filename
, line
, k
, "Failed to parse show status setting, ignoring: %s", rvalue
);
520 static int config_parse_crash_chvt(
522 const char *filename
,
525 unsigned section_line
,
538 r
= parse_crash_chvt(rvalue
);
540 log_syntax(unit
, LOG_ERR
, filename
, line
, r
, "Failed to parse CrashChangeVT= setting, ignoring: %s", rvalue
);
547 static int config_parse_join_controllers(const char *unit
,
548 const char *filename
,
551 unsigned section_line
,
558 const char *whole_rvalue
= rvalue
;
565 arg_join_controllers
= strv_free_free(arg_join_controllers
);
568 _cleanup_free_
char *word
= NULL
;
572 r
= extract_first_word(&rvalue
, &word
, WHITESPACE
, EXTRACT_QUOTES
);
574 log_syntax(unit
, LOG_ERR
, filename
, line
, r
, "Invalid value for %s: %s", lvalue
, whole_rvalue
);
580 l
= strv_split(word
, ",");
585 if (strv_length(l
) <= 1) {
590 if (!arg_join_controllers
) {
591 arg_join_controllers
= new(char**, 2);
592 if (!arg_join_controllers
) {
597 arg_join_controllers
[0] = l
;
598 arg_join_controllers
[1] = NULL
;
605 t
= new0(char**, n
+2);
613 for (a
= arg_join_controllers
; *a
; a
++) {
615 if (strv_overlap(*a
, l
)) {
616 if (strv_extend_strv(&l
, *a
, false) < 0) {
636 t
[n
++] = strv_uniq(l
);
638 strv_free_free(arg_join_controllers
);
639 arg_join_controllers
= t
;
642 if (!isempty(rvalue
))
643 log_syntax(unit
, LOG_ERR
, filename
, line
, 0, "Trailing garbage, ignoring.");
648 static int parse_config_file(void) {
650 const ConfigTableItem items
[] = {
651 { "Manager", "LogLevel", config_parse_level2
, 0, NULL
},
652 { "Manager", "LogTarget", config_parse_target
, 0, NULL
},
653 { "Manager", "LogColor", config_parse_color
, 0, NULL
},
654 { "Manager", "LogLocation", config_parse_location
, 0, NULL
},
655 { "Manager", "DumpCore", config_parse_bool
, 0, &arg_dump_core
},
656 { "Manager", "CrashChVT", /* legacy */ config_parse_crash_chvt
, 0, NULL
},
657 { "Manager", "CrashChangeVT", config_parse_crash_chvt
, 0, NULL
},
658 { "Manager", "CrashShell", config_parse_bool
, 0, &arg_crash_shell
},
659 { "Manager", "CrashReboot", config_parse_bool
, 0, &arg_crash_reboot
},
660 { "Manager", "ShowStatus", config_parse_show_status
, 0, &arg_show_status
},
661 { "Manager", "CPUAffinity", config_parse_cpu_affinity2
, 0, NULL
},
662 { "Manager", "JoinControllers", config_parse_join_controllers
, 0, &arg_join_controllers
},
663 { "Manager", "RuntimeWatchdogSec", config_parse_sec
, 0, &arg_runtime_watchdog
},
664 { "Manager", "ShutdownWatchdogSec", config_parse_sec
, 0, &arg_shutdown_watchdog
},
665 { "Manager", "CapabilityBoundingSet", config_parse_capability_set
, 0, &arg_capability_bounding_set
},
667 { "Manager", "SystemCallArchitectures", config_parse_syscall_archs
, 0, &arg_syscall_archs
},
669 { "Manager", "TimerSlackNSec", config_parse_nsec
, 0, &arg_timer_slack_nsec
},
670 { "Manager", "DefaultTimerAccuracySec", config_parse_sec
, 0, &arg_default_timer_accuracy_usec
},
671 { "Manager", "DefaultStandardOutput", config_parse_output
, 0, &arg_default_std_output
},
672 { "Manager", "DefaultStandardError", config_parse_output
, 0, &arg_default_std_error
},
673 { "Manager", "DefaultTimeoutStartSec", config_parse_sec
, 0, &arg_default_timeout_start_usec
},
674 { "Manager", "DefaultTimeoutStopSec", config_parse_sec
, 0, &arg_default_timeout_stop_usec
},
675 { "Manager", "DefaultRestartSec", config_parse_sec
, 0, &arg_default_restart_usec
},
676 { "Manager", "DefaultStartLimitInterval", config_parse_sec
, 0, &arg_default_start_limit_interval
},
677 { "Manager", "DefaultStartLimitBurst", config_parse_unsigned
, 0, &arg_default_start_limit_burst
},
678 { "Manager", "DefaultEnvironment", config_parse_environ
, 0, &arg_default_environment
},
679 { "Manager", "DefaultLimitCPU", config_parse_limit
, RLIMIT_CPU
, arg_default_rlimit
},
680 { "Manager", "DefaultLimitFSIZE", config_parse_limit
, RLIMIT_FSIZE
, arg_default_rlimit
},
681 { "Manager", "DefaultLimitDATA", config_parse_limit
, RLIMIT_DATA
, arg_default_rlimit
},
682 { "Manager", "DefaultLimitSTACK", config_parse_limit
, RLIMIT_STACK
, arg_default_rlimit
},
683 { "Manager", "DefaultLimitCORE", config_parse_limit
, RLIMIT_CORE
, arg_default_rlimit
},
684 { "Manager", "DefaultLimitRSS", config_parse_limit
, RLIMIT_RSS
, arg_default_rlimit
},
685 { "Manager", "DefaultLimitNOFILE", config_parse_limit
, RLIMIT_NOFILE
, arg_default_rlimit
},
686 { "Manager", "DefaultLimitAS", config_parse_limit
, RLIMIT_AS
, arg_default_rlimit
},
687 { "Manager", "DefaultLimitNPROC", config_parse_limit
, RLIMIT_NPROC
, arg_default_rlimit
},
688 { "Manager", "DefaultLimitMEMLOCK", config_parse_limit
, RLIMIT_MEMLOCK
, arg_default_rlimit
},
689 { "Manager", "DefaultLimitLOCKS", config_parse_limit
, RLIMIT_LOCKS
, arg_default_rlimit
},
690 { "Manager", "DefaultLimitSIGPENDING", config_parse_limit
, RLIMIT_SIGPENDING
, arg_default_rlimit
},
691 { "Manager", "DefaultLimitMSGQUEUE", config_parse_limit
, RLIMIT_MSGQUEUE
, arg_default_rlimit
},
692 { "Manager", "DefaultLimitNICE", config_parse_limit
, RLIMIT_NICE
, arg_default_rlimit
},
693 { "Manager", "DefaultLimitRTPRIO", config_parse_limit
, RLIMIT_RTPRIO
, arg_default_rlimit
},
694 { "Manager", "DefaultLimitRTTIME", config_parse_limit
, RLIMIT_RTTIME
, arg_default_rlimit
},
695 { "Manager", "DefaultCPUAccounting", config_parse_bool
, 0, &arg_default_cpu_accounting
},
696 { "Manager", "DefaultBlockIOAccounting", config_parse_bool
, 0, &arg_default_blockio_accounting
},
697 { "Manager", "DefaultMemoryAccounting", config_parse_bool
, 0, &arg_default_memory_accounting
},
698 { "Manager", "DefaultTasksAccounting", config_parse_bool
, 0, &arg_default_tasks_accounting
},
699 { "Manager", "DefaultTasksMax", config_parse_tasks_max
, 0, &arg_default_tasks_max
},
703 const char *fn
, *conf_dirs_nulstr
;
705 fn
= arg_running_as
== MANAGER_SYSTEM
?
706 PKGSYSCONFDIR
"/system.conf" :
707 PKGSYSCONFDIR
"/user.conf";
709 conf_dirs_nulstr
= arg_running_as
== MANAGER_SYSTEM
?
710 CONF_PATHS_NULSTR("systemd/system.conf.d") :
711 CONF_PATHS_NULSTR("systemd/user.conf.d");
713 config_parse_many(fn
, conf_dirs_nulstr
, "Manager\0", config_item_table_lookup
, items
, false, NULL
);
715 /* Traditionally "0" was used to turn off the default unit timeouts. Fix this up so that we used USEC_INFINITY
716 * like everywhere else. */
717 if (arg_default_timeout_start_usec
<= 0)
718 arg_default_timeout_start_usec
= USEC_INFINITY
;
719 if (arg_default_timeout_stop_usec
<= 0)
720 arg_default_timeout_stop_usec
= USEC_INFINITY
;
725 static void manager_set_defaults(Manager
*m
) {
729 m
->default_timer_accuracy_usec
= arg_default_timer_accuracy_usec
;
730 m
->default_std_output
= arg_default_std_output
;
731 m
->default_std_error
= arg_default_std_error
;
732 m
->default_timeout_start_usec
= arg_default_timeout_start_usec
;
733 m
->default_timeout_stop_usec
= arg_default_timeout_stop_usec
;
734 m
->default_restart_usec
= arg_default_restart_usec
;
735 m
->default_start_limit_interval
= arg_default_start_limit_interval
;
736 m
->default_start_limit_burst
= arg_default_start_limit_burst
;
737 m
->default_cpu_accounting
= arg_default_cpu_accounting
;
738 m
->default_blockio_accounting
= arg_default_blockio_accounting
;
739 m
->default_memory_accounting
= arg_default_memory_accounting
;
740 m
->default_tasks_accounting
= arg_default_tasks_accounting
;
741 m
->default_tasks_max
= arg_default_tasks_max
;
743 manager_set_default_rlimits(m
, arg_default_rlimit
);
744 manager_environment_add(m
, NULL
, arg_default_environment
);
747 static int parse_argv(int argc
, char *argv
[]) {
750 ARG_LOG_LEVEL
= 0x100,
760 ARG_DUMP_CONFIGURATION_ITEMS
,
769 ARG_DEFAULT_STD_OUTPUT
,
770 ARG_DEFAULT_STD_ERROR
,
774 static const struct option options
[] = {
775 { "log-level", required_argument
, NULL
, ARG_LOG_LEVEL
},
776 { "log-target", required_argument
, NULL
, ARG_LOG_TARGET
},
777 { "log-color", optional_argument
, NULL
, ARG_LOG_COLOR
},
778 { "log-location", optional_argument
, NULL
, ARG_LOG_LOCATION
},
779 { "unit", required_argument
, NULL
, ARG_UNIT
},
780 { "system", no_argument
, NULL
, ARG_SYSTEM
},
781 { "user", no_argument
, NULL
, ARG_USER
},
782 { "test", no_argument
, NULL
, ARG_TEST
},
783 { "no-pager", no_argument
, NULL
, ARG_NO_PAGER
},
784 { "help", no_argument
, NULL
, 'h' },
785 { "version", no_argument
, NULL
, ARG_VERSION
},
786 { "dump-configuration-items", no_argument
, NULL
, ARG_DUMP_CONFIGURATION_ITEMS
},
787 { "dump-core", optional_argument
, NULL
, ARG_DUMP_CORE
},
788 { "crash-chvt", required_argument
, NULL
, ARG_CRASH_CHVT
},
789 { "crash-shell", optional_argument
, NULL
, ARG_CRASH_SHELL
},
790 { "crash-reboot", optional_argument
, NULL
, ARG_CRASH_REBOOT
},
791 { "confirm-spawn", optional_argument
, NULL
, ARG_CONFIRM_SPAWN
},
792 { "show-status", optional_argument
, NULL
, ARG_SHOW_STATUS
},
793 { "deserialize", required_argument
, NULL
, ARG_DESERIALIZE
},
794 { "switched-root", no_argument
, NULL
, ARG_SWITCHED_ROOT
},
795 { "default-standard-output", required_argument
, NULL
, ARG_DEFAULT_STD_OUTPUT
, },
796 { "default-standard-error", required_argument
, NULL
, ARG_DEFAULT_STD_ERROR
, },
797 { "machine-id", required_argument
, NULL
, ARG_MACHINE_ID
},
809 while ((c
= getopt_long(argc
, argv
, "hDbsz:", options
, NULL
)) >= 0)
814 r
= log_set_max_level_from_string(optarg
);
816 log_error("Failed to parse log level %s.", optarg
);
823 r
= log_set_target_from_string(optarg
);
825 log_error("Failed to parse log target %s.", optarg
);
834 r
= log_show_color_from_string(optarg
);
836 log_error("Failed to parse log color setting %s.", optarg
);
840 log_show_color(true);
844 case ARG_LOG_LOCATION
:
846 r
= log_show_location_from_string(optarg
);
848 log_error("Failed to parse log location setting %s.", optarg
);
852 log_show_location(true);
856 case ARG_DEFAULT_STD_OUTPUT
:
857 r
= exec_output_from_string(optarg
);
859 log_error("Failed to parse default standard output setting %s.", optarg
);
862 arg_default_std_output
= r
;
865 case ARG_DEFAULT_STD_ERROR
:
866 r
= exec_output_from_string(optarg
);
868 log_error("Failed to parse default standard error output setting %s.", optarg
);
871 arg_default_std_error
= r
;
876 r
= free_and_strdup(&arg_default_unit
, optarg
);
878 return log_error_errno(r
, "Failed to set default unit %s: %m", optarg
);
883 arg_running_as
= MANAGER_SYSTEM
;
887 arg_running_as
= MANAGER_USER
;
891 arg_action
= ACTION_TEST
;
892 if (arg_no_pager
< 0)
901 arg_action
= ACTION_VERSION
;
904 case ARG_DUMP_CONFIGURATION_ITEMS
:
905 arg_action
= ACTION_DUMP_CONFIGURATION_ITEMS
;
910 arg_dump_core
= true;
912 r
= parse_boolean(optarg
);
914 return log_error_errno(r
, "Failed to parse dump core boolean: %s", optarg
);
920 r
= parse_crash_chvt(optarg
);
922 return log_error_errno(r
, "Failed to parse crash virtual terminal index: %s", optarg
);
925 case ARG_CRASH_SHELL
:
927 arg_crash_shell
= true;
929 r
= parse_boolean(optarg
);
931 return log_error_errno(r
, "Failed to parse crash shell boolean: %s", optarg
);
936 case ARG_CRASH_REBOOT
:
938 arg_crash_reboot
= true;
940 r
= parse_boolean(optarg
);
942 return log_error_errno(r
, "Failed to parse crash shell boolean: %s", optarg
);
943 arg_crash_reboot
= r
;
947 case ARG_CONFIRM_SPAWN
:
948 r
= optarg
? parse_boolean(optarg
) : 1;
950 log_error("Failed to parse confirm spawn boolean %s.", optarg
);
953 arg_confirm_spawn
= r
;
956 case ARG_SHOW_STATUS
:
958 r
= parse_show_status(optarg
, &arg_show_status
);
960 log_error("Failed to parse show status boolean %s.", optarg
);
964 arg_show_status
= SHOW_STATUS_YES
;
967 case ARG_DESERIALIZE
: {
971 r
= safe_atoi(optarg
, &fd
);
972 if (r
< 0 || fd
< 0) {
973 log_error("Failed to parse deserialize option %s.", optarg
);
977 (void) fd_cloexec(fd
, true);
981 return log_error_errno(errno
, "Failed to open serialization fd: %m");
983 safe_fclose(arg_serialization
);
984 arg_serialization
= f
;
989 case ARG_SWITCHED_ROOT
:
990 arg_switched_root
= true;
994 r
= set_machine_id(optarg
);
996 log_error("MachineID '%s' is not valid.", optarg
);
1002 arg_action
= ACTION_HELP
;
1003 if (arg_no_pager
< 0)
1004 arg_no_pager
= true;
1008 log_set_max_level(LOG_DEBUG
);
1014 /* Just to eat away the sysvinit kernel
1015 * cmdline args without getopt() error
1016 * messages that we'll parse in
1017 * parse_proc_cmdline_word() or ignore. */
1026 assert_not_reached("Unhandled option code.");
1029 if (optind
< argc
&& getpid() != 1) {
1030 /* Hmm, when we aren't run as init system
1031 * let's complain about excess arguments */
1033 log_error("Excess arguments.");
1040 static int help(void) {
1042 printf("%s [OPTIONS...]\n\n"
1043 "Starts up and maintains the system or user services.\n\n"
1044 " -h --help Show this help\n"
1045 " --test Determine startup sequence, dump it and exit\n"
1046 " --no-pager Do not pipe output into a pager\n"
1047 " --dump-configuration-items Dump understood unit configuration items\n"
1048 " --unit=UNIT Set default unit\n"
1049 " --system Run a system instance, even if PID != 1\n"
1050 " --user Run a user instance\n"
1051 " --dump-core[=BOOL] Dump core on crash\n"
1052 " --crash-vt=NR Change to specified VT on crash\n"
1053 " --crash-reboot[=BOOL] Reboot on crash\n"
1054 " --crash-shell[=BOOL] Run shell on crash\n"
1055 " --confirm-spawn[=BOOL] Ask for confirmation when spawning processes\n"
1056 " --show-status[=BOOL] Show status updates on the console during bootup\n"
1057 " --log-target=TARGET Set log target (console, journal, kmsg, journal-or-kmsg, null)\n"
1058 " --log-level=LEVEL Set log level (debug, info, notice, warning, err, crit, alert, emerg)\n"
1059 " --log-color[=BOOL] Highlight important log messages\n"
1060 " --log-location[=BOOL] Include code location in log messages\n"
1061 " --default-standard-output= Set default standard output for services\n"
1062 " --default-standard-error= Set default standard error output for services\n",
1063 program_invocation_short_name
);
1068 static int prepare_reexecute(Manager
*m
, FILE **_f
, FDSet
**_fds
, bool switching_root
) {
1069 _cleanup_fdset_free_ FDSet
*fds
= NULL
;
1070 _cleanup_fclose_
FILE *f
= NULL
;
1077 r
= manager_open_serialization(m
, &f
);
1079 return log_error_errno(r
, "Failed to create serialization file: %m");
1081 /* Make sure nothing is really destructed when we shut down */
1083 bus_manager_send_reloading(m
, true);
1089 r
= manager_serialize(m
, f
, fds
, switching_root
);
1091 return log_error_errno(r
, "Failed to serialize state: %m");
1093 if (fseeko(f
, 0, SEEK_SET
) == (off_t
) -1)
1094 return log_error_errno(errno
, "Failed to rewind serialization fd: %m");
1096 r
= fd_cloexec(fileno(f
), false);
1098 return log_error_errno(r
, "Failed to disable O_CLOEXEC for serialization: %m");
1100 r
= fdset_cloexec(fds
, false);
1102 return log_error_errno(r
, "Failed to disable O_CLOEXEC for serialization fds: %m");
1113 static int bump_rlimit_nofile(struct rlimit
*saved_rlimit
) {
1117 assert(saved_rlimit
);
1119 /* Save the original RLIMIT_NOFILE so that we can reset it
1120 * later when transitioning from the initrd to the main
1121 * systemd or suchlike. */
1122 if (getrlimit(RLIMIT_NOFILE
, saved_rlimit
) < 0)
1123 return log_error_errno(errno
, "Reading RLIMIT_NOFILE failed: %m");
1125 /* Make sure forked processes get the default kernel setting */
1126 if (!arg_default_rlimit
[RLIMIT_NOFILE
]) {
1129 rl
= newdup(struct rlimit
, saved_rlimit
, 1);
1133 arg_default_rlimit
[RLIMIT_NOFILE
] = rl
;
1136 /* Bump up the resource limit for ourselves substantially */
1137 nl
.rlim_cur
= nl
.rlim_max
= 64*1024;
1138 r
= setrlimit_closest(RLIMIT_NOFILE
, &nl
);
1140 return log_error_errno(r
, "Setting RLIMIT_NOFILE failed: %m");
1145 static void test_usr(void) {
1147 /* Check that /usr is not a separate fs */
1149 if (dir_is_empty("/usr") <= 0)
1152 log_warning("/usr appears to be on its own filesystem and is not already mounted. This is not a supported setup. "
1153 "Some things will probably break (sometimes even silently) in mysterious ways. "
1154 "Consult http://freedesktop.org/wiki/Software/systemd/separate-usr-is-broken for more information.");
1157 static int initialize_join_controllers(void) {
1158 /* By default, mount "cpu" + "cpuacct" together, and "net_cls"
1159 * + "net_prio". We'd like to add "cpuset" to the mix, but
1160 * "cpuset" doesn't really work for groups with no initialized
1163 arg_join_controllers
= new(char**, 3);
1164 if (!arg_join_controllers
)
1167 arg_join_controllers
[0] = strv_new("cpu", "cpuacct", NULL
);
1168 if (!arg_join_controllers
[0])
1171 arg_join_controllers
[1] = strv_new("net_cls", "net_prio", NULL
);
1172 if (!arg_join_controllers
[1])
1175 arg_join_controllers
[2] = NULL
;
1179 arg_join_controllers
= strv_free_free(arg_join_controllers
);
1183 static int enforce_syscall_archs(Set
*archs
) {
1185 scmp_filter_ctx
*seccomp
;
1190 seccomp
= seccomp_init(SCMP_ACT_ALLOW
);
1194 SET_FOREACH(id
, arg_syscall_archs
, i
) {
1195 r
= seccomp_arch_add(seccomp
, PTR_TO_UINT32(id
) - 1);
1199 log_error_errno(r
, "Failed to add architecture to seccomp: %m");
1204 r
= seccomp_attr_set(seccomp
, SCMP_FLTATR_CTL_NNP
, 0);
1206 log_error_errno(r
, "Failed to unset NO_NEW_PRIVS: %m");
1210 r
= seccomp_load(seccomp
);
1212 log_error_errno(r
, "Failed to add install architecture seccomp: %m");
1215 seccomp_release(seccomp
);
1222 static int status_welcome(void) {
1223 _cleanup_free_
char *pretty_name
= NULL
, *ansi_color
= NULL
;
1226 r
= parse_env_file("/etc/os-release", NEWLINE
,
1227 "PRETTY_NAME", &pretty_name
,
1228 "ANSI_COLOR", &ansi_color
,
1231 r
= parse_env_file("/usr/lib/os-release", NEWLINE
,
1232 "PRETTY_NAME", &pretty_name
,
1233 "ANSI_COLOR", &ansi_color
,
1236 if (r
< 0 && r
!= -ENOENT
)
1237 log_warning_errno(r
, "Failed to read os-release file: %m");
1239 return status_printf(NULL
, false, false,
1240 "\nWelcome to \x1B[%sm%s\x1B[0m!\n",
1241 isempty(ansi_color
) ? "1" : ansi_color
,
1242 isempty(pretty_name
) ? "Linux" : pretty_name
);
1245 static int write_container_id(void) {
1249 c
= getenv("container");
1253 r
= write_string_file("/run/systemd/container", c
, WRITE_STRING_FILE_CREATE
);
1255 return log_warning_errno(r
, "Failed to write /run/systemd/container, ignoring: %m");
1260 static int bump_unix_max_dgram_qlen(void) {
1261 _cleanup_free_
char *qlen
= NULL
;
1265 /* Let's bump the net.unix.max_dgram_qlen sysctl. The kernel
1266 * default of 16 is simply too low. We set the value really
1267 * really early during boot, so that it is actually applied to
1268 * all our sockets, including the $NOTIFY_SOCKET one. */
1270 r
= read_one_line_file("/proc/sys/net/unix/max_dgram_qlen", &qlen
);
1272 return log_warning_errno(r
, "Failed to read AF_UNIX datagram queue length, ignoring: %m");
1274 r
= safe_atolu(qlen
, &v
);
1276 return log_warning_errno(r
, "Failed to parse AF_UNIX datagram queue length, ignoring: %m");
1278 if (v
>= DEFAULT_UNIX_MAX_DGRAM_QLEN
)
1282 if (asprintf(&qlen
, "%lu\n", DEFAULT_UNIX_MAX_DGRAM_QLEN
) < 0)
1285 r
= write_string_file("/proc/sys/net/unix/max_dgram_qlen", qlen
, 0);
1287 return log_full_errno(IN_SET(r
, -EROFS
, -EPERM
, -EACCES
) ? LOG_DEBUG
: LOG_WARNING
, r
,
1288 "Failed to bump AF_UNIX datagram queue length, ignoring: %m");
1293 int main(int argc
, char *argv
[]) {
1295 int r
, retval
= EXIT_FAILURE
;
1296 usec_t before_startup
, after_startup
;
1297 char timespan
[FORMAT_TIMESPAN_MAX
];
1299 bool reexecute
= false;
1300 const char *shutdown_verb
= NULL
;
1301 dual_timestamp initrd_timestamp
= DUAL_TIMESTAMP_NULL
;
1302 dual_timestamp userspace_timestamp
= DUAL_TIMESTAMP_NULL
;
1303 dual_timestamp kernel_timestamp
= DUAL_TIMESTAMP_NULL
;
1304 dual_timestamp security_start_timestamp
= DUAL_TIMESTAMP_NULL
;
1305 dual_timestamp security_finish_timestamp
= DUAL_TIMESTAMP_NULL
;
1306 static char systemd
[] = "systemd";
1307 bool skip_setup
= false;
1309 bool loaded_policy
= false;
1310 bool arm_reboot_watchdog
= false;
1311 bool queue_default_job
= false;
1312 bool empty_etc
= false;
1313 char *switch_root_dir
= NULL
, *switch_root_init
= NULL
;
1314 struct rlimit saved_rlimit_nofile
= RLIMIT_MAKE_CONST(0);
1315 const char *error_message
= NULL
;
1317 #ifdef HAVE_SYSV_COMPAT
1318 if (getpid() != 1 && strstr(program_invocation_short_name
, "init")) {
1319 /* This is compatibility support for SysV, where
1320 * calling init as a user is identical to telinit. */
1323 execv(SYSTEMCTL_BINARY_PATH
, argv
);
1324 log_error_errno(errno
, "Failed to exec " SYSTEMCTL_BINARY_PATH
": %m");
1329 dual_timestamp_from_monotonic(&kernel_timestamp
, 0);
1330 dual_timestamp_get(&userspace_timestamp
);
1332 /* Determine if this is a reexecution or normal bootup. We do
1333 * the full command line parsing much later, so let's just
1334 * have a quick peek here. */
1335 if (strv_find(argv
+1, "--deserialize"))
1338 /* If we have switched root, do all the special setup
1340 if (strv_find(argv
+1, "--switched-root"))
1343 /* If we get started via the /sbin/init symlink then we are
1344 called 'init'. After a subsequent reexecution we are then
1345 called 'systemd'. That is confusing, hence let's call us
1346 systemd right-away. */
1347 program_invocation_short_name
= systemd
;
1348 prctl(PR_SET_NAME
, systemd
);
1353 log_show_color(isatty(STDERR_FILENO
) > 0);
1354 log_set_upgrade_syslog_to_journal(true);
1356 /* Disable the umask logic */
1360 if (getpid() == 1 && detect_container() <= 0) {
1362 /* Running outside of a container as PID 1 */
1363 arg_running_as
= MANAGER_SYSTEM
;
1365 log_set_target(LOG_TARGET_KMSG
);
1369 initrd_timestamp
= userspace_timestamp
;
1372 r
= mount_setup_early();
1374 error_message
= "Failed to early mount API filesystems";
1377 dual_timestamp_get(&security_start_timestamp
);
1378 if (mac_selinux_setup(&loaded_policy
) < 0) {
1379 error_message
= "Failed to load SELinux policy";
1381 } else if (ima_setup() < 0) {
1382 error_message
= "Failed to load IMA policy";
1384 } else if (mac_smack_setup(&loaded_policy
) < 0) {
1385 error_message
= "Failed to load SMACK policy";
1388 dual_timestamp_get(&security_finish_timestamp
);
1391 if (mac_selinux_init(NULL
) < 0) {
1392 error_message
= "Failed to initialize SELinux policy";
1397 if (clock_is_localtime() > 0) {
1401 * The very first call of settimeofday() also does a time warp in the kernel.
1403 * In the rtc-in-local time mode, we set the kernel's timezone, and rely on
1404 * external tools to take care of maintaining the RTC and do all adjustments.
1405 * This matches the behavior of Windows, which leaves the RTC alone if the
1406 * registry tells that the RTC runs in UTC.
1408 r
= clock_set_timezone(&min
);
1410 log_error_errno(r
, "Failed to apply local time delta, ignoring: %m");
1412 log_info("RTC configured in localtime, applying delta of %i minutes to system time.", min
);
1413 } else if (!in_initrd()) {
1415 * Do a dummy very first call to seal the kernel's time warp magic.
1417 * Do not call this this from inside the initrd. The initrd might not
1418 * carry /etc/adjtime with LOCAL, but the real system could be set up
1419 * that way. In such case, we need to delay the time-warp or the sealing
1420 * until we reach the real system.
1422 * Do no set the kernel's timezone. The concept of local time cannot
1423 * be supported reliably, the time will jump or be incorrect at every daylight
1424 * saving time change. All kernel local time concepts will be treated
1427 clock_reset_timewarp();
1431 /* Set the default for later on, but don't actually
1432 * open the logs like this for now. Note that if we
1433 * are transitioning from the initrd there might still
1434 * be journal fd open, and we shouldn't attempt
1435 * opening that before we parsed /proc/cmdline which
1436 * might redirect output elsewhere. */
1437 log_set_target(LOG_TARGET_JOURNAL_OR_KMSG
);
1439 } else if (getpid() == 1) {
1440 /* Running inside a container, as PID 1 */
1441 arg_running_as
= MANAGER_SYSTEM
;
1442 log_set_target(LOG_TARGET_CONSOLE
);
1443 log_close_console(); /* force reopen of /dev/console */
1446 /* For the later on, see above... */
1447 log_set_target(LOG_TARGET_JOURNAL
);
1449 /* clear the kernel timestamp,
1450 * because we are in a container */
1451 kernel_timestamp
.monotonic
= 0ULL;
1452 kernel_timestamp
.realtime
= 0ULL;
1455 /* Running as user instance */
1456 arg_running_as
= MANAGER_USER
;
1457 log_set_target(LOG_TARGET_AUTO
);
1460 /* clear the kernel timestamp,
1461 * because we are not PID 1 */
1462 kernel_timestamp
= DUAL_TIMESTAMP_NULL
;
1465 /* Initialize default unit */
1466 r
= free_and_strdup(&arg_default_unit
, SPECIAL_DEFAULT_TARGET
);
1468 log_emergency_errno(r
, "Failed to set default unit %s: %m", SPECIAL_DEFAULT_TARGET
);
1469 error_message
= "Failed to set default unit";
1473 r
= initialize_join_controllers();
1475 error_message
= "Failed to initialize cgroup controllers";
1479 /* Mount /proc, /sys and friends, so that /proc/cmdline and
1480 * /proc/$PID/fd is available. */
1481 if (getpid() == 1) {
1483 /* Load the kernel modules early, so that we kdbus.ko is loaded before kdbusfs shall be mounted */
1487 r
= mount_setup(loaded_policy
);
1489 error_message
= "Failed to mount API filesystems";
1494 /* Reset all signal handlers. */
1495 (void) reset_all_signal_handlers();
1496 (void) ignore_signals(SIGNALS_IGNORE
, -1);
1498 if (parse_config_file() < 0) {
1499 error_message
= "Failed to parse config file";
1503 if (arg_running_as
== MANAGER_SYSTEM
) {
1504 r
= parse_proc_cmdline(parse_proc_cmdline_item
);
1506 log_warning_errno(r
, "Failed to parse kernel command line, ignoring: %m");
1509 /* Note that this also parses bits from the kernel command
1510 * line, including "debug". */
1511 log_parse_environment();
1513 if (parse_argv(argc
, argv
) < 0) {
1514 error_message
= "Failed to parse commandline arguments";
1518 if (arg_action
== ACTION_TEST
&&
1520 log_error("Don't run test mode as root.");
1524 if (arg_running_as
== MANAGER_USER
&&
1525 arg_action
== ACTION_RUN
&&
1527 log_error("Trying to run as user instance, but the system has not been booted with systemd.");
1531 if (arg_running_as
== MANAGER_SYSTEM
&&
1532 arg_action
== ACTION_RUN
&&
1533 running_in_chroot() > 0) {
1534 log_error("Cannot be run in a chroot() environment.");
1538 if (arg_action
== ACTION_TEST
)
1541 pager_open_if_enabled();
1543 if (arg_action
== ACTION_HELP
) {
1546 } else if (arg_action
== ACTION_VERSION
) {
1549 } else if (arg_action
== ACTION_DUMP_CONFIGURATION_ITEMS
) {
1550 unit_dump_config_items(stdout
);
1551 retval
= EXIT_SUCCESS
;
1553 } else if (arg_action
== ACTION_DONE
) {
1554 retval
= EXIT_SUCCESS
;
1558 if (arg_running_as
== MANAGER_USER
&&
1559 !getenv("XDG_RUNTIME_DIR")) {
1560 log_error("Trying to run as user instance, but $XDG_RUNTIME_DIR is not set.");
1564 assert_se(arg_action
== ACTION_RUN
|| arg_action
== ACTION_TEST
);
1566 /* Close logging fds, in order not to confuse fdset below */
1569 /* Remember open file descriptors for later deserialization */
1570 r
= fdset_new_fill(&fds
);
1572 log_emergency_errno(r
, "Failed to allocate fd set: %m");
1573 error_message
= "Failed to allocate fd set";
1576 fdset_cloexec(fds
, true);
1578 if (arg_serialization
)
1579 assert_se(fdset_remove(fds
, fileno(arg_serialization
)) >= 0);
1581 if (arg_running_as
== MANAGER_SYSTEM
)
1582 /* Become a session leader if we aren't one yet. */
1585 /* Move out of the way, so that we won't block unmounts */
1586 assert_se(chdir("/") == 0);
1588 /* Reset the console, but only if this is really init and we
1589 * are freshly booted */
1590 if (arg_running_as
== MANAGER_SYSTEM
&& arg_action
== ACTION_RUN
) {
1592 /* If we are init, we connect stdin/stdout/stderr to
1593 * /dev/null and make sure we don't have a controlling
1597 if (getpid() == 1 && !skip_setup
)
1601 /* Open the logging devices, if possible and necessary */
1604 if (arg_show_status
== _SHOW_STATUS_UNSET
)
1605 arg_show_status
= SHOW_STATUS_YES
;
1607 /* Make sure we leave a core dump without panicing the
1609 if (getpid() == 1) {
1610 install_crash_handler();
1612 r
= mount_cgroup_controllers(arg_join_controllers
);
1617 if (arg_running_as
== MANAGER_SYSTEM
) {
1620 log_info(PACKAGE_STRING
" running in %ssystem mode. (" SYSTEMD_FEATURES
")",
1621 arg_action
== ACTION_TEST
? "test " : "" );
1623 v
= detect_virtualization();
1625 log_info("Detected virtualization %s.", virtualization_to_string(v
));
1627 write_container_id();
1629 log_info("Detected architecture %s.", architecture_to_string(uname_architecture()));
1632 log_info("Running in initial RAM disk.");
1634 /* Let's check whether /etc is already populated. We
1635 * don't actually really check for that, but use
1636 * /etc/machine-id as flag file. This allows container
1637 * managers and installers to provision a couple of
1638 * files already. If the container manager wants to
1639 * provision the machine ID itself it should pass
1640 * $container_uuid to PID 1. */
1642 empty_etc
= access("/etc/machine-id", F_OK
) < 0;
1644 log_info("Running with unpopulated /etc.");
1646 _cleanup_free_
char *t
;
1648 t
= uid_to_name(getuid());
1649 log_debug(PACKAGE_STRING
" running in %suser mode for user "UID_FMT
"/%s. (" SYSTEMD_FEATURES
")",
1650 arg_action
== ACTION_TEST
? " test" : "", getuid(), t
);
1653 if (arg_running_as
== MANAGER_SYSTEM
&& !skip_setup
) {
1654 if (arg_show_status
> 0)
1658 machine_id_setup(NULL
, arg_machine_id
);
1660 bump_unix_max_dgram_qlen();
1665 if (arg_running_as
== MANAGER_SYSTEM
&& arg_runtime_watchdog
> 0)
1666 watchdog_set_timeout(&arg_runtime_watchdog
);
1668 if (arg_timer_slack_nsec
!= NSEC_INFINITY
)
1669 if (prctl(PR_SET_TIMERSLACK
, arg_timer_slack_nsec
) < 0)
1670 log_error_errno(errno
, "Failed to adjust timer slack: %m");
1672 if (!cap_test_all(arg_capability_bounding_set
)) {
1673 r
= capability_bounding_set_drop_usermode(arg_capability_bounding_set
);
1675 log_emergency_errno(r
, "Failed to drop capability bounding set of usermode helpers: %m");
1676 error_message
= "Failed to drop capability bounding set of usermode helpers";
1679 r
= capability_bounding_set_drop(arg_capability_bounding_set
, true);
1681 log_emergency_errno(r
, "Failed to drop capability bounding set: %m");
1682 error_message
= "Failed to drop capability bounding set";
1687 if (arg_syscall_archs
) {
1688 r
= enforce_syscall_archs(arg_syscall_archs
);
1690 error_message
= "Failed to set syscall architectures";
1695 if (arg_running_as
== MANAGER_USER
)
1696 /* Become reaper of our children */
1697 if (prctl(PR_SET_CHILD_SUBREAPER
, 1) < 0)
1698 log_warning_errno(errno
, "Failed to make us a subreaper: %m");
1700 if (arg_running_as
== MANAGER_SYSTEM
) {
1701 bump_rlimit_nofile(&saved_rlimit_nofile
);
1704 r
= unit_file_preset_all(UNIT_FILE_SYSTEM
, false, NULL
, UNIT_FILE_PRESET_ENABLE_ONLY
, false, NULL
, 0);
1706 log_full_errno(r
== -EEXIST
? LOG_NOTICE
: LOG_WARNING
, r
, "Failed to populate /etc with preset unit settings, ignoring: %m");
1708 log_info("Populated /etc with preset unit settings.");
1712 r
= manager_new(arg_running_as
, arg_action
== ACTION_TEST
, &m
);
1714 log_emergency_errno(r
, "Failed to allocate manager object: %m");
1715 error_message
= "Failed to allocate manager object";
1719 m
->confirm_spawn
= arg_confirm_spawn
;
1720 m
->runtime_watchdog
= arg_runtime_watchdog
;
1721 m
->shutdown_watchdog
= arg_shutdown_watchdog
;
1722 m
->userspace_timestamp
= userspace_timestamp
;
1723 m
->kernel_timestamp
= kernel_timestamp
;
1724 m
->initrd_timestamp
= initrd_timestamp
;
1725 m
->security_start_timestamp
= security_start_timestamp
;
1726 m
->security_finish_timestamp
= security_finish_timestamp
;
1728 manager_set_defaults(m
);
1729 manager_set_show_status(m
, arg_show_status
);
1730 manager_set_first_boot(m
, empty_etc
);
1732 /* Remember whether we should queue the default job */
1733 queue_default_job
= !arg_serialization
|| arg_switched_root
;
1735 before_startup
= now(CLOCK_MONOTONIC
);
1737 r
= manager_startup(m
, arg_serialization
, fds
);
1739 log_error_errno(r
, "Failed to fully start up daemon: %m");
1741 /* This will close all file descriptors that were opened, but
1742 * not claimed by any unit. */
1743 fds
= fdset_free(fds
);
1745 arg_serialization
= safe_fclose(arg_serialization
);
1747 if (queue_default_job
) {
1748 _cleanup_(sd_bus_error_free
) sd_bus_error error
= SD_BUS_ERROR_NULL
;
1749 Unit
*target
= NULL
;
1750 Job
*default_unit_job
;
1752 log_debug("Activating default unit: %s", arg_default_unit
);
1754 r
= manager_load_unit(m
, arg_default_unit
, NULL
, &error
, &target
);
1756 log_error("Failed to load default target: %s", bus_error_message(&error
, r
));
1757 else if (target
->load_state
== UNIT_ERROR
|| target
->load_state
== UNIT_NOT_FOUND
)
1758 log_error_errno(target
->load_error
, "Failed to load default target: %m");
1759 else if (target
->load_state
== UNIT_MASKED
)
1760 log_error("Default target masked.");
1762 if (!target
|| target
->load_state
!= UNIT_LOADED
) {
1763 log_info("Trying to load rescue target...");
1765 r
= manager_load_unit(m
, SPECIAL_RESCUE_TARGET
, NULL
, &error
, &target
);
1767 log_emergency("Failed to load rescue target: %s", bus_error_message(&error
, r
));
1768 error_message
= "Failed to load rescue target";
1770 } else if (target
->load_state
== UNIT_ERROR
|| target
->load_state
== UNIT_NOT_FOUND
) {
1771 log_emergency_errno(target
->load_error
, "Failed to load rescue target: %m");
1772 error_message
= "Failed to load rescue target";
1774 } else if (target
->load_state
== UNIT_MASKED
) {
1775 log_emergency("Rescue target masked.");
1776 error_message
= "Rescue target masked";
1781 assert(target
->load_state
== UNIT_LOADED
);
1783 if (arg_action
== ACTION_TEST
) {
1784 printf("-> By units:\n");
1785 manager_dump_units(m
, stdout
, "\t");
1788 r
= manager_add_job(m
, JOB_START
, target
, JOB_ISOLATE
, &error
, &default_unit_job
);
1790 log_debug("Default target could not be isolated, starting instead: %s", bus_error_message(&error
, r
));
1792 sd_bus_error_free(&error
);
1794 r
= manager_add_job(m
, JOB_START
, target
, JOB_REPLACE
, &error
, &default_unit_job
);
1796 log_emergency("Failed to start default target: %s", bus_error_message(&error
, r
));
1797 error_message
= "Failed to start default target";
1801 log_emergency("Failed to isolate default target: %s", bus_error_message(&error
, r
));
1802 error_message
= "Failed to isolate default target";
1806 m
->default_unit_job_id
= default_unit_job
->id
;
1808 after_startup
= now(CLOCK_MONOTONIC
);
1809 log_full(arg_action
== ACTION_TEST
? LOG_INFO
: LOG_DEBUG
,
1810 "Loaded units and determined initial transaction in %s.",
1811 format_timespan(timespan
, sizeof(timespan
), after_startup
- before_startup
, 100 * USEC_PER_MSEC
));
1813 if (arg_action
== ACTION_TEST
) {
1814 printf("-> By jobs:\n");
1815 manager_dump_jobs(m
, stdout
, "\t");
1816 retval
= EXIT_SUCCESS
;
1822 r
= manager_loop(m
);
1824 log_emergency_errno(r
, "Failed to run main loop: %m");
1825 error_message
= "Failed to run main loop";
1829 switch (m
->exit_code
) {
1831 case MANAGER_RELOAD
:
1832 log_info("Reloading.");
1834 r
= parse_config_file();
1836 log_error("Failed to parse config file.");
1838 manager_set_defaults(m
);
1840 r
= manager_reload(m
);
1842 log_error_errno(r
, "Failed to reload: %m");
1845 case MANAGER_REEXECUTE
:
1847 if (prepare_reexecute(m
, &arg_serialization
, &fds
, false) < 0) {
1848 error_message
= "Failed to prepare for reexecution";
1853 log_notice("Reexecuting.");
1856 case MANAGER_SWITCH_ROOT
:
1857 /* Steal the switch root parameters */
1858 switch_root_dir
= m
->switch_root
;
1859 switch_root_init
= m
->switch_root_init
;
1860 m
->switch_root
= m
->switch_root_init
= NULL
;
1862 if (!switch_root_init
)
1863 if (prepare_reexecute(m
, &arg_serialization
, &fds
, true) < 0) {
1864 error_message
= "Failed to prepare for reexecution";
1869 log_notice("Switching root.");
1873 retval
= m
->return_value
;
1875 if (m
->running_as
== MANAGER_USER
) {
1881 case MANAGER_REBOOT
:
1882 case MANAGER_POWEROFF
:
1884 case MANAGER_KEXEC
: {
1885 static const char * const table
[_MANAGER_EXIT_CODE_MAX
] = {
1886 [MANAGER_EXIT
] = "exit",
1887 [MANAGER_REBOOT
] = "reboot",
1888 [MANAGER_POWEROFF
] = "poweroff",
1889 [MANAGER_HALT
] = "halt",
1890 [MANAGER_KEXEC
] = "kexec"
1893 assert_se(shutdown_verb
= table
[m
->exit_code
]);
1894 arm_reboot_watchdog
= m
->exit_code
== MANAGER_REBOOT
;
1896 log_notice("Shutting down.");
1901 assert_not_reached("Unknown exit code.");
1909 arg_shutdown_watchdog
= m
->shutdown_watchdog
;
1911 m
= manager_free(m
);
1913 for (j
= 0; j
< ELEMENTSOF(arg_default_rlimit
); j
++)
1914 arg_default_rlimit
[j
] = mfree(arg_default_rlimit
[j
]);
1916 arg_default_unit
= mfree(arg_default_unit
);
1917 arg_join_controllers
= strv_free_free(arg_join_controllers
);
1918 arg_default_environment
= strv_free(arg_default_environment
);
1919 arg_syscall_archs
= set_free(arg_syscall_archs
);
1921 mac_selinux_finish();
1925 unsigned i
, args_size
;
1927 /* Close and disarm the watchdog, so that the new
1928 * instance can reinitialize it, but doesn't get
1929 * rebooted while we do that */
1930 watchdog_close(true);
1932 /* Reset the RLIMIT_NOFILE to the kernel default, so
1933 * that the new systemd can pass the kernel default to
1934 * its child processes */
1935 if (saved_rlimit_nofile
.rlim_cur
> 0)
1936 (void) setrlimit(RLIMIT_NOFILE
, &saved_rlimit_nofile
);
1938 if (switch_root_dir
) {
1939 /* Kill all remaining processes from the
1940 * initrd, but don't wait for them, so that we
1941 * can handle the SIGCHLD for them after
1943 broadcast_signal(SIGTERM
, false, true);
1945 /* And switch root with MS_MOVE, because we remove the old directory afterwards and detach it. */
1946 r
= switch_root(switch_root_dir
, "/mnt", true, MS_MOVE
);
1948 log_error_errno(r
, "Failed to switch root, trying to continue: %m");
1951 args_size
= MAX(6, argc
+1);
1952 args
= newa(const char*, args_size
);
1954 if (!switch_root_init
) {
1955 char sfd
[DECIMAL_STR_MAX(int) + 1];
1957 /* First try to spawn ourselves with the right
1958 * path, and with full serialization. We do
1959 * this only if the user didn't specify an
1960 * explicit init to spawn. */
1962 assert(arg_serialization
);
1965 xsprintf(sfd
, "%i", fileno(arg_serialization
));
1968 args
[i
++] = SYSTEMD_BINARY_PATH
;
1969 if (switch_root_dir
)
1970 args
[i
++] = "--switched-root";
1971 args
[i
++] = arg_running_as
== MANAGER_SYSTEM
? "--system" : "--user";
1972 args
[i
++] = "--deserialize";
1976 /* do not pass along the environment we inherit from the kernel or initrd */
1977 if (switch_root_dir
)
1980 assert(i
<= args_size
);
1983 * We want valgrind to print its memory usage summary before reexecution.
1984 * Valgrind won't do this is on its own on exec(), but it will do it on exit().
1985 * Hence, to ensure we get a summary here, fork() off a child, let it exit() cleanly,
1986 * so that it prints the summary, and wait() for it in the parent, before proceeding into the exec().
1988 valgrind_summary_hack();
1990 (void) execv(args
[0], (char* const*) args
);
1993 /* Try the fallback, if there is any, without any
1994 * serialization. We pass the original argv[] and
1995 * envp[]. (Well, modulo the ordering changes due to
1996 * getopt() in argv[], and some cleanups in envp[],
1997 * but let's hope that doesn't matter.) */
1999 arg_serialization
= safe_fclose(arg_serialization
);
2000 fds
= fdset_free(fds
);
2002 /* Reopen the console */
2003 (void) make_console_stdio();
2005 for (j
= 1, i
= 1; j
< (unsigned) argc
; j
++)
2006 args
[i
++] = argv
[j
];
2008 assert(i
<= args_size
);
2010 /* Reenable any blocked signals, especially important
2011 * if we switch from initial ramdisk to init=... */
2012 (void) reset_all_signal_handlers();
2013 (void) reset_signal_mask();
2015 if (switch_root_init
) {
2016 args
[0] = switch_root_init
;
2017 (void) execv(args
[0], (char* const*) args
);
2018 log_warning_errno(errno
, "Failed to execute configured init, trying fallback: %m");
2021 args
[0] = "/sbin/init";
2022 (void) execv(args
[0], (char* const*) args
);
2024 if (errno
== ENOENT
) {
2025 log_warning("No /sbin/init, trying fallback");
2027 args
[0] = "/bin/sh";
2029 (void) execv(args
[0], (char* const*) args
);
2030 log_error_errno(errno
, "Failed to execute /bin/sh, giving up: %m");
2032 log_warning_errno(errno
, "Failed to execute /sbin/init, giving up: %m");
2035 arg_serialization
= safe_fclose(arg_serialization
);
2036 fds
= fdset_free(fds
);
2038 #ifdef HAVE_VALGRIND_VALGRIND_H
2039 /* If we are PID 1 and running under valgrind, then let's exit
2040 * here explicitly. valgrind will only generate nice output on
2041 * exit(), not on exec(), hence let's do the former not the
2043 if (getpid() == 1 && RUNNING_ON_VALGRIND
)
2047 if (shutdown_verb
) {
2048 char log_level
[DECIMAL_STR_MAX(int) + 1];
2049 char exit_code
[DECIMAL_STR_MAX(uint8_t) + 1];
2050 const char* command_line
[11] = {
2051 SYSTEMD_SHUTDOWN_BINARY_PATH
,
2053 "--log-level", log_level
,
2057 _cleanup_strv_free_
char **env_block
= NULL
;
2059 assert(command_line
[pos
] == NULL
);
2060 env_block
= strv_copy(environ
);
2062 xsprintf(log_level
, "%d", log_get_max_level());
2064 switch (log_get_target()) {
2066 case LOG_TARGET_KMSG
:
2067 case LOG_TARGET_JOURNAL_OR_KMSG
:
2068 case LOG_TARGET_SYSLOG_OR_KMSG
:
2069 command_line
[pos
++] = "kmsg";
2072 case LOG_TARGET_NULL
:
2073 command_line
[pos
++] = "null";
2076 case LOG_TARGET_CONSOLE
:
2078 command_line
[pos
++] = "console";
2082 if (log_get_show_color())
2083 command_line
[pos
++] = "--log-color";
2085 if (log_get_show_location())
2086 command_line
[pos
++] = "--log-location";
2088 if (streq(shutdown_verb
, "exit")) {
2089 command_line
[pos
++] = "--exit-code";
2090 command_line
[pos
++] = exit_code
;
2091 xsprintf(exit_code
, "%d", retval
);
2094 assert(pos
< ELEMENTSOF(command_line
));
2096 if (arm_reboot_watchdog
&& arg_shutdown_watchdog
> 0) {
2099 /* If we reboot let's set the shutdown
2100 * watchdog and tell the shutdown binary to
2101 * repeatedly ping it */
2102 r
= watchdog_set_timeout(&arg_shutdown_watchdog
);
2103 watchdog_close(r
< 0);
2105 /* Tell the binary how often to ping, ignore failure */
2106 if (asprintf(&e
, "WATCHDOG_USEC="USEC_FMT
, arg_shutdown_watchdog
) > 0)
2107 (void) strv_push(&env_block
, e
);
2109 watchdog_close(true);
2111 /* Avoid the creation of new processes forked by the
2112 * kernel; at this point, we will not listen to the
2114 if (detect_container() <= 0)
2115 (void) cg_uninstall_release_agent(SYSTEMD_CGROUP_CONTROLLER
);
2117 execve(SYSTEMD_SHUTDOWN_BINARY_PATH
, (char **) command_line
, env_block
);
2118 log_error_errno(errno
, "Failed to execute shutdown binary, %s: %m",
2119 getpid() == 1 ? "freezing" : "quitting");
2122 if (getpid() == 1) {
2124 manager_status_printf(NULL
, STATUS_TYPE_EMERGENCY
,
2125 ANSI_HIGHLIGHT_RED
"!!!!!!" ANSI_NORMAL
,
2126 "%s, freezing.", error_message
);