]> git.ipfire.org Git - thirdparty/systemd.git/blame - src/core/main.c
tree-wide: uniformly bump RLIMIT_NOFILE in all our tools that access the journal
[thirdparty/systemd.git] / src / core / main.c
CommitLineData
53e1b683 1/* SPDX-License-Identifier: LGPL-2.1+ */
a7334b09 2
60918275 3#include <errno.h>
3dfc9763 4#include <fcntl.h>
f170852a 5#include <getopt.h>
97c4f35c 6#include <signal.h>
3dfc9763
LP
7#include <stdio.h>
8#include <string.h>
664f88a7 9#include <sys/mount.h>
3dfc9763 10#include <sys/prctl.h>
b9e74c39 11#include <sys/reboot.h>
3dfc9763
LP
12#include <sys/stat.h>
13#include <unistd.h>
349cc4a5 14#if HAVE_SECCOMP
b64a3d86
LP
15#include <seccomp.h>
16#endif
349cc4a5 17#if HAVE_VALGRIND_VALGRIND_H
3dfc9763
LP
18#include <valgrind/valgrind.h>
19#endif
54b434b1 20
718db961 21#include "sd-bus.h"
cf0fbc49 22#include "sd-daemon.h"
b2e7486c 23#include "sd-messages.h"
3dfc9763 24
b5efdb8a 25#include "alloc-util.h"
d9d93745 26#include "architecture.h"
3dfc9763
LP
27#include "build.h"
28#include "bus-error.h"
29#include "bus-util.h"
430f0182 30#include "capability-util.h"
24efb112 31#include "clock-util.h"
3dfc9763 32#include "conf-parser.h"
618234a5 33#include "cpu-set-util.h"
bbc1acab 34#include "dbus.h"
3dfc9763
LP
35#include "dbus-manager.h"
36#include "def.h"
eee8b7ab 37#include "emergency-action.h"
3dfc9763 38#include "env-util.h"
57b7a260 39#include "exit-status.h"
3ffd4af2 40#include "fd-util.h"
3dfc9763 41#include "fdset.h"
718db961 42#include "fileio.h"
f97b34a6 43#include "format-util.h"
f4f15635 44#include "fs-util.h"
3dfc9763
LP
45#include "hostname-setup.h"
46#include "ima-setup.h"
47#include "killall.h"
48#include "kmod-setup.h"
d7b8eec7 49#include "load-fragment.h"
3dfc9763 50#include "log.h"
b6e66135 51#include "loopback-setup.h"
b6e66135 52#include "machine-id-setup.h"
3dfc9763
LP
53#include "manager.h"
54#include "missing.h"
55#include "mount-setup.h"
d58ad743 56#include "os-util.h"
3dfc9763 57#include "pager.h"
6bedfcbb 58#include "parse-util.h"
7d5ceb64 59#include "path-util.h"
4e731273 60#include "proc-cmdline.h"
3dfc9763 61#include "process-util.h"
8869a0b4 62#include "raw-clone.h"
78f22b97 63#include "rlimit-util.h"
349cc4a5 64#if HAVE_SECCOMP
83f12b27
FS
65#include "seccomp-util.h"
66#endif
b6e66135 67#include "selinux-setup.h"
3dfc9763
LP
68#include "selinux-util.h"
69#include "signal-util.h"
ffbd2c4d 70#include "smack-setup.h"
3dfc9763 71#include "special.h"
8fcde012 72#include "stat-util.h"
15a5e950 73#include "stdio-util.h"
3dfc9763
LP
74#include "strv.h"
75#include "switch-root.h"
76#include "terminal-util.h"
8612da97 77#include "umask-util.h"
b1d4f8e1 78#include "user-util.h"
9ce17593 79#include "util.h"
3dfc9763
LP
80#include "virt.h"
81#include "watchdog.h"
b6e66135 82
f170852a
LP
83static enum {
84 ACTION_RUN,
e965d56d 85 ACTION_HELP,
9ba0bc4e 86 ACTION_VERSION,
e537352b 87 ACTION_TEST,
bbc1acab
YW
88 ACTION_DUMP_CONFIGURATION_ITEMS,
89 ACTION_DUMP_BUS_PROPERTIES,
fa0f4d8a 90} arg_action = ACTION_RUN;
fa0f4d8a 91static char *arg_default_unit = NULL;
463d0d15 92static bool arg_system = false;
fa0f4d8a 93static bool arg_dump_core = true;
fa0f4d8a 94static int arg_crash_chvt = -1;
b9e74c39
LP
95static bool arg_crash_shell = false;
96static bool arg_crash_reboot = false;
7d5ceb64 97static char *arg_confirm_spawn = NULL;
7a293242 98static ShowStatus arg_show_status = _SHOW_STATUS_INVALID;
bf4df7c3 99static bool arg_switched_root = false;
ea4b98e6 100static bool arg_no_pager = false;
2a12e32e 101static bool arg_service_watchdogs = true;
0c85a4f3 102static char ***arg_join_controllers = NULL;
706343f4 103static ExecOutput arg_default_std_output = EXEC_OUTPUT_JOURNAL;
0a494f1f 104static ExecOutput arg_default_std_error = EXEC_OUTPUT_INHERIT;
1f19a534
OS
105static usec_t arg_default_restart_usec = DEFAULT_RESTART_USEC;
106static usec_t arg_default_timeout_start_usec = DEFAULT_TIMEOUT_USEC;
107static usec_t arg_default_timeout_stop_usec = DEFAULT_TIMEOUT_USEC;
3f41e1e5
LN
108static usec_t arg_default_start_limit_interval = DEFAULT_START_LIMIT_INTERVAL;
109static unsigned arg_default_start_limit_burst = DEFAULT_START_LIMIT_BURST;
e96d6be7
LP
110static usec_t arg_runtime_watchdog = 0;
111static usec_t arg_shutdown_watchdog = 10 * USEC_PER_MINUTE;
c6885f5f 112static char *arg_early_core_pattern = NULL;
8a2c1fbf 113static char *arg_watchdog_device = NULL;
97d0e5f8 114static char **arg_default_environment = NULL;
517d56b1 115static struct rlimit *arg_default_rlimit[_RLIMIT_MAX] = {};
a103496c 116static uint64_t arg_capability_bounding_set = CAP_ALL;
39362f6f 117static bool arg_no_new_privs = false;
3a43da28 118static nsec_t arg_timer_slack_nsec = NSEC_INFINITY;
bd8f585b 119static usec_t arg_default_timer_accuracy_usec = 1 * USEC_PER_MINUTE;
d3b1c508
LP
120static Set* arg_syscall_archs = NULL;
121static FILE* arg_serialization = NULL;
085afe36 122static bool arg_default_cpu_accounting = false;
13c31542 123static bool arg_default_io_accounting = false;
377bfd2d 124static bool arg_default_ip_accounting = false;
085afe36 125static bool arg_default_blockio_accounting = false;
444d5863 126static bool arg_default_memory_accounting = MEMORY_ACCOUNTING_DEFAULT;
9ded9cd1 127static bool arg_default_tasks_accounting = true;
79baeeb9 128static uint64_t arg_default_tasks_max = UINT64_MAX;
ee48dbd5 129static sd_id128_t arg_machine_id = {};
ae8c7939 130static EmergencyAction arg_cad_burst_action = EMERGENCY_ACTION_REBOOT_FORCE;
4fc935ca 131
848e863a 132_noreturn_ static void freeze_or_reboot(void) {
b9e74c39
LP
133
134 if (arg_crash_reboot) {
135 log_notice("Rebooting in 10s...");
136 (void) sleep(10);
137
138 log_notice("Rebooting now...");
139 (void) reboot(RB_AUTOBOOT);
140 log_emergency_errno(errno, "Failed to reboot: %m");
141 }
142
143 log_emergency("Freezing execution.");
144 freeze();
145}
146
848e863a 147_noreturn_ static void crash(int sig) {
7d06dad9
MS
148 struct sigaction sa;
149 pid_t pid;
97c4f35c 150
df0ff127 151 if (getpid_cached() != 1)
abb26902 152 /* Pass this on immediately, if this is not PID 1 */
92ca4cac 153 (void) raise(sig);
abb26902 154 else if (!arg_dump_core)
4104970e 155 log_emergency("Caught <%s>, not dumping core.", signal_to_string(sig));
97c4f35c 156 else {
7d06dad9 157 sa = (struct sigaction) {
189d5bac 158 .sa_handler = nop_signal_handler,
b92bea5d
ZJS
159 .sa_flags = SA_NOCLDSTOP|SA_RESTART,
160 };
97c4f35c 161
6f5e3f35 162 /* We want to wait for the core process, hence let's enable SIGCHLD */
92ca4cac 163 (void) sigaction(SIGCHLD, &sa, NULL);
6f5e3f35 164
8869a0b4 165 pid = raw_clone(SIGCHLD);
e62d8c39 166 if (pid < 0)
56f64d95 167 log_emergency_errno(errno, "Caught <%s>, cannot fork for core dump: %m", signal_to_string(sig));
97c4f35c 168 else if (pid == 0) {
97c4f35c 169 /* Enable default signal handler for core dump */
15a90032 170
92ca4cac
LP
171 sa = (struct sigaction) {
172 .sa_handler = SIG_DFL,
173 };
174 (void) sigaction(sig, &sa, NULL);
97c4f35c 175
15a90032
LP
176 /* Don't limit the coredump size */
177 (void) setrlimit(RLIMIT_CORE, &RLIMIT_MAKE_CONST(RLIM_INFINITY));
97c4f35c
LP
178
179 /* Just to be sure... */
e62d9b81 180 (void) chdir("/");
97c4f35c
LP
181
182 /* Raise the signal again */
ee05e779 183 pid = raw_getpid();
92ca4cac 184 (void) kill(pid, sig); /* raise() would kill the parent */
97c4f35c
LP
185
186 assert_not_reached("We shouldn't be here...");
92ca4cac 187 _exit(EXIT_FAILURE);
4fc935ca 188 } else {
8e12a6ae
LP
189 siginfo_t status;
190 int r;
4fc935ca
LP
191
192 /* Order things nicely. */
e62d8c39
ZJS
193 r = wait_for_terminate(pid, &status);
194 if (r < 0)
da927ba9 195 log_emergency_errno(r, "Caught <%s>, waitpid() failed: %m", signal_to_string(sig));
8e12a6ae 196 else if (status.si_code != CLD_DUMPED)
ee05e779
ZJS
197 log_emergency("Caught <%s>, core dump failed (child "PID_FMT", code=%s, status=%i/%s).",
198 signal_to_string(sig),
199 pid, sigchld_code_to_string(status.si_code),
200 status.si_status,
201 strna(status.si_code == CLD_EXITED
38107f5a 202 ? exit_status_to_string(status.si_status, EXIT_STATUS_MINIMAL)
ee05e779 203 : signal_to_string(status.si_status)));
4fc935ca 204 else
4104970e 205 log_emergency("Caught <%s>, dumped core as pid "PID_FMT".", signal_to_string(sig), pid);
97c4f35c
LP
206 }
207 }
208
b9e74c39 209 if (arg_crash_chvt >= 0)
92ca4cac 210 (void) chvt(arg_crash_chvt);
601f6a1e 211
7d06dad9
MS
212 sa = (struct sigaction) {
213 .sa_handler = SIG_IGN,
214 .sa_flags = SA_NOCLDSTOP|SA_NOCLDWAIT|SA_RESTART,
215 };
216
217 /* Let the kernel reap children for us */
218 (void) sigaction(SIGCHLD, &sa, NULL);
8c43883a 219
7d06dad9 220 if (arg_crash_shell) {
b9e74c39 221 log_notice("Executing crash shell in 10s...");
92ca4cac 222 (void) sleep(10);
4fc935ca 223
8869a0b4 224 pid = raw_clone(SIGCHLD);
cd3bd60a 225 if (pid < 0)
56f64d95 226 log_emergency_errno(errno, "Failed to fork off crash shell: %m");
6f5e3f35 227 else if (pid == 0) {
b9e74c39 228 (void) setsid();
92ca4cac
LP
229 (void) make_console_stdio();
230 (void) execle("/bin/sh", "/bin/sh", NULL, environ);
6f5e3f35 231
ee05e779 232 log_emergency_errno(errno, "execle() failed: %m");
92ca4cac 233 _exit(EXIT_FAILURE);
b9e74c39
LP
234 } else {
235 log_info("Spawned crash shell as PID "PID_FMT".", pid);
4cf0b03b 236 (void) wait_for_terminate(pid, NULL);
b9e74c39 237 }
4fc935ca
LP
238 }
239
b9e74c39 240 freeze_or_reboot();
97c4f35c
LP
241}
242
243static void install_crash_handler(void) {
297d563d 244 static const struct sigaction sa = {
b92bea5d 245 .sa_handler = crash,
297d563d 246 .sa_flags = SA_NODEFER, /* So that we can raise the signal again from the signal handler */
b92bea5d 247 };
297d563d 248 int r;
97c4f35c 249
297d563d
LP
250 /* We ignore the return value here, since, we don't mind if we
251 * cannot set up a crash handler */
252 r = sigaction_many(&sa, SIGNALS_CRASH_HANDLER, -1);
253 if (r < 0)
254 log_debug_errno(r, "I had trouble setting up the crash handler, ignoring: %m");
97c4f35c 255}
f170852a 256
56d96fc0
LP
257static int console_setup(void) {
258 _cleanup_close_ int tty_fd = -1;
259 int r;
80876c20 260
512947d4 261 tty_fd = open_terminal("/dev/console", O_WRONLY|O_NOCTTY|O_CLOEXEC);
23bbb0de
MS
262 if (tty_fd < 0)
263 return log_error_errno(tty_fd, "Failed to open /dev/console: %m");
80876c20 264
56d96fc0
LP
265 /* We don't want to force text mode. plymouth may be showing
266 * pictures already from initrd. */
512947d4 267 r = reset_terminal_fd(tty_fd, false);
23bbb0de
MS
268 if (r < 0)
269 return log_error_errno(r, "Failed to reset /dev/console: %m");
843d2643 270
56d96fc0 271 return 0;
80876c20
LP
272}
273
b9e74c39
LP
274static int parse_crash_chvt(const char *value) {
275 int b;
276
277 if (safe_atoi(value, &arg_crash_chvt) >= 0)
278 return 0;
279
280 b = parse_boolean(value);
281 if (b < 0)
282 return b;
283
284 if (b > 0)
285 arg_crash_chvt = 0; /* switch to where kmsg goes */
286 else
287 arg_crash_chvt = -1; /* turn off switching */
288
289 return 0;
290}
291
7d5ceb64
FB
292static int parse_confirm_spawn(const char *value, char **console) {
293 char *s;
294 int r;
295
296 r = value ? parse_boolean(value) : 1;
297 if (r == 0) {
298 *console = NULL;
299 return 0;
300 }
301
302 if (r > 0) /* on with default tty */
303 s = strdup("/dev/console");
304 else if (is_path(value)) /* on with fully qualified path */
305 s = strdup(value);
306 else /* on with only a tty file name, not a fully qualified path */
307 s = strjoin("/dev/", value);
308 if (!s)
309 return -ENOMEM;
5b65ae15 310
7d5ceb64
FB
311 *console = s;
312 return 0;
313}
314
ee48dbd5 315static int set_machine_id(const char *m) {
e042eab7 316 sd_id128_t t;
8b26cdbd 317 assert(m);
ee48dbd5 318
e042eab7 319 if (sd_id128_from_string(m, &t) < 0)
ee48dbd5
NC
320 return -EINVAL;
321
e042eab7 322 if (sd_id128_is_null(t))
ee48dbd5
NC
323 return -EINVAL;
324
e042eab7 325 arg_machine_id = t;
ee48dbd5
NC
326 return 0;
327}
328
96287a49 329static int parse_proc_cmdline_item(const char *key, const char *value, void *data) {
f170852a 330
059cb385 331 int r;
f170852a 332
059cb385 333 assert(key);
5192bd19 334
1d84ad94 335 if (STR_IN_SET(key, "systemd.unit", "rd.systemd.unit")) {
bf4df7c3 336
1d84ad94
LP
337 if (proc_cmdline_value_missing(key, value))
338 return 0;
bf4df7c3 339
1d84ad94
LP
340 if (!unit_name_is_valid(value, UNIT_NAME_PLAIN|UNIT_NAME_INSTANCE))
341 log_warning("Unit name specified on %s= is not valid, ignoring: %s", key, value);
342 else if (in_initrd() == !!startswith(key, "rd.")) {
343 if (free_and_strdup(&arg_default_unit, value) < 0)
344 return log_oom();
345 }
f170852a 346
1d84ad94 347 } else if (proc_cmdline_key_streq(key, "systemd.dump_core")) {
4fc935ca 348
1d84ad94 349 r = value ? parse_boolean(value) : true;
fb472900 350 if (r < 0)
059cb385 351 log_warning("Failed to parse dump core switch %s. Ignoring.", value);
4fc935ca 352 else
fa0f4d8a 353 arg_dump_core = r;
4fc935ca 354
c6885f5f
FB
355 } else if (proc_cmdline_key_streq(key, "systemd.early_core_pattern")) {
356
357 if (proc_cmdline_value_missing(key, value))
358 return 0;
359
360 if (path_is_absolute(value))
361 (void) parse_path_argument_and_warn(value, false, &arg_early_core_pattern);
362 else
363 log_warning("Specified core pattern '%s' is not an absolute path, ignoring.", value);
364
1d84ad94 365 } else if (proc_cmdline_key_streq(key, "systemd.crash_chvt")) {
b9e74c39 366
1d84ad94
LP
367 if (!value)
368 arg_crash_chvt = 0; /* turn on */
369 else if (parse_crash_chvt(value) < 0)
b9e74c39
LP
370 log_warning("Failed to parse crash chvt switch %s. Ignoring.", value);
371
1d84ad94 372 } else if (proc_cmdline_key_streq(key, "systemd.crash_shell")) {
4fc935ca 373
1d84ad94 374 r = value ? parse_boolean(value) : true;
fb472900 375 if (r < 0)
059cb385 376 log_warning("Failed to parse crash shell switch %s. Ignoring.", value);
4fc935ca 377 else
fa0f4d8a 378 arg_crash_shell = r;
5e7ee61c 379
1d84ad94 380 } else if (proc_cmdline_key_streq(key, "systemd.crash_reboot")) {
5e7ee61c 381
1d84ad94 382 r = value ? parse_boolean(value) : true;
b9e74c39
LP
383 if (r < 0)
384 log_warning("Failed to parse crash reboot switch %s. Ignoring.", value);
5e7ee61c 385 else
b9e74c39 386 arg_crash_reboot = r;
5e7ee61c 387
1d84ad94
LP
388 } else if (proc_cmdline_key_streq(key, "systemd.confirm_spawn")) {
389 char *s;
7d5ceb64 390
1d84ad94 391 r = parse_confirm_spawn(value, &s);
059cb385 392 if (r < 0)
7d5ceb64 393 log_warning_errno(r, "Failed to parse confirm_spawn switch %s. Ignoring.", value);
1d84ad94
LP
394 else {
395 free(arg_confirm_spawn);
396 arg_confirm_spawn = s;
397 }
601f6a1e 398
2a12e32e
JK
399 } else if (proc_cmdline_key_streq(key, "systemd.service_watchdogs")) {
400
401 r = value ? parse_boolean(value) : true;
402 if (r < 0)
403 log_warning("Failed to parse service watchdog switch %s. Ignoring.", value);
404 else
405 arg_service_watchdogs = r;
406
1d84ad94 407 } else if (proc_cmdline_key_streq(key, "systemd.show_status")) {
9e58ff9c 408
1d84ad94
LP
409 if (value) {
410 r = parse_show_status(value, &arg_show_status);
411 if (r < 0)
412 log_warning("Failed to parse show status switch %s. Ignoring.", value);
413 } else
414 arg_show_status = SHOW_STATUS_YES;
059cb385 415
1d84ad94
LP
416 } else if (proc_cmdline_key_streq(key, "systemd.default_standard_output")) {
417
418 if (proc_cmdline_value_missing(key, value))
419 return 0;
0a494f1f 420
059cb385 421 r = exec_output_from_string(value);
fb472900 422 if (r < 0)
059cb385 423 log_warning("Failed to parse default standard output switch %s. Ignoring.", value);
0a494f1f
LP
424 else
425 arg_default_std_output = r;
0a494f1f 426
1d84ad94
LP
427 } else if (proc_cmdline_key_streq(key, "systemd.default_standard_error")) {
428
429 if (proc_cmdline_value_missing(key, value))
430 return 0;
059cb385
LP
431
432 r = exec_output_from_string(value);
fb472900 433 if (r < 0)
059cb385 434 log_warning("Failed to parse default standard error switch %s. Ignoring.", value);
0a494f1f
LP
435 else
436 arg_default_std_error = r;
9e7c5357 437
1d84ad94
LP
438 } else if (streq(key, "systemd.setenv")) {
439
440 if (proc_cmdline_value_missing(key, value))
441 return 0;
059cb385
LP
442
443 if (env_assignment_is_valid(value)) {
e21fea24
KS
444 char **env;
445
059cb385 446 env = strv_env_set(arg_default_environment, value);
1d84ad94
LP
447 if (!env)
448 return log_oom();
449
450 arg_default_environment = env;
e21fea24 451 } else
059cb385 452 log_warning("Environment variable name '%s' is not valid. Ignoring.", value);
9e58ff9c 453
1d84ad94
LP
454 } else if (proc_cmdline_key_streq(key, "systemd.machine_id")) {
455
456 if (proc_cmdline_value_missing(key, value))
457 return 0;
458
459 r = set_machine_id(value);
460 if (r < 0)
461 log_warning("MachineID '%s' is not valid. Ignoring.", value);
ee48dbd5 462
1d84ad94
LP
463 } else if (proc_cmdline_key_streq(key, "systemd.default_timeout_start_sec")) {
464
465 if (proc_cmdline_value_missing(key, value))
466 return 0;
467
468 r = parse_sec(value, &arg_default_timeout_start_usec);
469 if (r < 0)
470 log_warning_errno(r, "Failed to parse default start timeout: %s, ignoring.", value);
471
472 if (arg_default_timeout_start_usec <= 0)
473 arg_default_timeout_start_usec = USEC_INFINITY;
ee48dbd5 474
8a2c1fbf
EJ
475 } else if (proc_cmdline_key_streq(key, "systemd.watchdog_device")) {
476
477 if (proc_cmdline_value_missing(key, value))
478 return 0;
479
c6885f5f 480 (void) parse_path_argument_and_warn(value, false, &arg_watchdog_device);
8a2c1fbf 481
059cb385 482 } else if (streq(key, "quiet") && !value) {
d7b15e0a 483
7a293242 484 if (arg_show_status == _SHOW_STATUS_INVALID)
d450b6f2 485 arg_show_status = SHOW_STATUS_AUTO;
059cb385
LP
486
487 } else if (streq(key, "debug") && !value) {
d7b15e0a 488
1de1c9c3
LP
489 /* Note that log_parse_environment() handles 'debug'
490 * too, and sets the log level to LOG_DEBUG. */
d7b15e0a 491
75f86906 492 if (detect_container() > 0)
b2103dcc 493 log_set_target(LOG_TARGET_CONSOLE);
059cb385 494
dcd61450 495 } else if (!value) {
e2c9a131 496 const char *target;
f170852a
LP
497
498 /* SysV compatibility */
e2c9a131
EV
499 target = runlevel_to_target(key);
500 if (target)
501 return free_and_strdup(&arg_default_unit, target);
f170852a
LP
502 }
503
504 return 0;
505}
506
e8e581bf
ZJS
507#define DEFINE_SETTER(name, func, descr) \
508 static int name(const char *unit, \
509 const char *filename, \
510 unsigned line, \
511 const char *section, \
71a61510 512 unsigned section_line, \
e8e581bf
ZJS
513 const char *lvalue, \
514 int ltype, \
515 const char *rvalue, \
516 void *data, \
517 void *userdata) { \
518 \
519 int r; \
520 \
521 assert(filename); \
522 assert(lvalue); \
523 assert(rvalue); \
524 \
525 r = func(rvalue); \
526 if (r < 0) \
d1cefe0a
LP
527 log_syntax(unit, LOG_ERR, filename, line, r, \
528 "Invalid " descr "'%s': %m", \
529 rvalue); \
e8e581bf
ZJS
530 \
531 return 0; \
532 }
487393e9 533
a6ecbf83
FB
534DEFINE_SETTER(config_parse_level2, log_set_max_level_from_string, "log level");
535DEFINE_SETTER(config_parse_target, log_set_target_from_string, "target");
536DEFINE_SETTER(config_parse_color, log_show_color_from_string, "color" );
537DEFINE_SETTER(config_parse_location, log_show_location_from_string, "location");
487393e9 538
4d7213b2
LP
539static int config_parse_cpu_affinity2(
540 const char *unit,
541 const char *filename,
542 unsigned line,
543 const char *section,
544 unsigned section_line,
545 const char *lvalue,
546 int ltype,
547 const char *rvalue,
548 void *data,
549 void *userdata) {
487393e9 550
4b40bc38 551 _cleanup_cpu_free_ cpu_set_t *c = NULL;
f5c72b73 552 int ncpus;
487393e9 553
765d143b 554 ncpus = parse_cpu_set_and_warn(rvalue, &c, unit, filename, line, lvalue);
f5c72b73
FB
555 if (ncpus < 0)
556 return ncpus;
487393e9 557
f5c72b73 558 if (sched_setaffinity(0, CPU_ALLOC_SIZE(ncpus), c) < 0)
25f027c5 559 log_warning_errno(errno, "Failed to set CPU affinity: %m");
487393e9 560
487393e9
LP
561 return 0;
562}
563
4d7213b2
LP
564static int config_parse_show_status(
565 const char* unit,
566 const char *filename,
567 unsigned line,
568 const char *section,
569 unsigned section_line,
570 const char *lvalue,
571 int ltype,
572 const char *rvalue,
573 void *data,
574 void *userdata) {
575
576 int k;
577 ShowStatus *b = data;
578
579 assert(filename);
580 assert(lvalue);
581 assert(rvalue);
582 assert(data);
583
584 k = parse_show_status(rvalue, b);
585 if (k < 0) {
12ca818f 586 log_syntax(unit, LOG_ERR, filename, line, k, "Failed to parse show status setting, ignoring: %s", rvalue);
4d7213b2
LP
587 return 0;
588 }
589
590 return 0;
591}
592
5db98187
LP
593static int config_parse_output_restricted(
594 const char* unit,
595 const char *filename,
596 unsigned line,
597 const char *section,
598 unsigned section_line,
599 const char *lvalue,
600 int ltype,
601 const char *rvalue,
602 void *data,
603 void *userdata) {
604
605 ExecOutput t, *eo = data;
606
607 assert(filename);
608 assert(lvalue);
609 assert(rvalue);
610 assert(data);
611
612 t = exec_output_from_string(rvalue);
613 if (t < 0) {
614 log_syntax(unit, LOG_ERR, filename, line, 0, "Failed to parse output type, ignoring: %s", rvalue);
615 return 0;
616 }
617
566b7d23
ZD
618 if (IN_SET(t, EXEC_OUTPUT_SOCKET, EXEC_OUTPUT_NAMED_FD, EXEC_OUTPUT_FILE, EXEC_OUTPUT_FILE_APPEND)) {
619 log_syntax(unit, LOG_ERR, filename, line, 0, "Standard output types socket, fd:, file:, append: are not supported as defaults, ignoring: %s", rvalue);
5db98187
LP
620 return 0;
621 }
622
623 *eo = t;
624 return 0;
625}
626
b9e74c39
LP
627static int config_parse_crash_chvt(
628 const char* unit,
629 const char *filename,
630 unsigned line,
631 const char *section,
632 unsigned section_line,
633 const char *lvalue,
634 int ltype,
635 const char *rvalue,
636 void *data,
637 void *userdata) {
638
639 int r;
640
641 assert(filename);
642 assert(lvalue);
643 assert(rvalue);
b9e74c39
LP
644
645 r = parse_crash_chvt(rvalue);
646 if (r < 0) {
647 log_syntax(unit, LOG_ERR, filename, line, r, "Failed to parse CrashChangeVT= setting, ignoring: %s", rvalue);
648 return 0;
649 }
650
651 return 0;
652}
653
487393e9
LP
654static int parse_config_file(void) {
655
f975e971 656 const ConfigTableItem items[] = {
d3b1c508
LP
657 { "Manager", "LogLevel", config_parse_level2, 0, NULL },
658 { "Manager", "LogTarget", config_parse_target, 0, NULL },
659 { "Manager", "LogColor", config_parse_color, 0, NULL },
660 { "Manager", "LogLocation", config_parse_location, 0, NULL },
661 { "Manager", "DumpCore", config_parse_bool, 0, &arg_dump_core },
b9e74c39
LP
662 { "Manager", "CrashChVT", /* legacy */ config_parse_crash_chvt, 0, NULL },
663 { "Manager", "CrashChangeVT", config_parse_crash_chvt, 0, NULL },
d3b1c508 664 { "Manager", "CrashShell", config_parse_bool, 0, &arg_crash_shell },
b9e74c39 665 { "Manager", "CrashReboot", config_parse_bool, 0, &arg_crash_reboot },
d3b1c508 666 { "Manager", "ShowStatus", config_parse_show_status, 0, &arg_show_status },
d3b1c508
LP
667 { "Manager", "CPUAffinity", config_parse_cpu_affinity2, 0, NULL },
668 { "Manager", "JoinControllers", config_parse_join_controllers, 0, &arg_join_controllers },
669 { "Manager", "RuntimeWatchdogSec", config_parse_sec, 0, &arg_runtime_watchdog },
670 { "Manager", "ShutdownWatchdogSec", config_parse_sec, 0, &arg_shutdown_watchdog },
8a2c1fbf 671 { "Manager", "WatchdogDevice", config_parse_path, 0, &arg_watchdog_device },
a103496c 672 { "Manager", "CapabilityBoundingSet", config_parse_capability_set, 0, &arg_capability_bounding_set },
39362f6f 673 { "Manager", "NoNewPrivileges", config_parse_bool, 0, &arg_no_new_privs },
349cc4a5 674#if HAVE_SECCOMP
d3b1c508 675 { "Manager", "SystemCallArchitectures", config_parse_syscall_archs, 0, &arg_syscall_archs },
89fffa27 676#endif
d3b1c508 677 { "Manager", "TimerSlackNSec", config_parse_nsec, 0, &arg_timer_slack_nsec },
bd8f585b 678 { "Manager", "DefaultTimerAccuracySec", config_parse_sec, 0, &arg_default_timer_accuracy_usec },
5db98187
LP
679 { "Manager", "DefaultStandardOutput", config_parse_output_restricted,0, &arg_default_std_output },
680 { "Manager", "DefaultStandardError", config_parse_output_restricted,0, &arg_default_std_error },
d3b1c508
LP
681 { "Manager", "DefaultTimeoutStartSec", config_parse_sec, 0, &arg_default_timeout_start_usec },
682 { "Manager", "DefaultTimeoutStopSec", config_parse_sec, 0, &arg_default_timeout_stop_usec },
683 { "Manager", "DefaultRestartSec", config_parse_sec, 0, &arg_default_restart_usec },
f0367da7
LP
684 { "Manager", "DefaultStartLimitInterval", config_parse_sec, 0, &arg_default_start_limit_interval }, /* obsolete alias */
685 { "Manager", "DefaultStartLimitIntervalSec",config_parse_sec, 0, &arg_default_start_limit_interval },
d3b1c508
LP
686 { "Manager", "DefaultStartLimitBurst", config_parse_unsigned, 0, &arg_default_start_limit_burst },
687 { "Manager", "DefaultEnvironment", config_parse_environ, 0, &arg_default_environment },
4f424df7
LP
688 { "Manager", "DefaultLimitCPU", config_parse_rlimit, RLIMIT_CPU, arg_default_rlimit },
689 { "Manager", "DefaultLimitFSIZE", config_parse_rlimit, RLIMIT_FSIZE, arg_default_rlimit },
690 { "Manager", "DefaultLimitDATA", config_parse_rlimit, RLIMIT_DATA, arg_default_rlimit },
691 { "Manager", "DefaultLimitSTACK", config_parse_rlimit, RLIMIT_STACK, arg_default_rlimit },
692 { "Manager", "DefaultLimitCORE", config_parse_rlimit, RLIMIT_CORE, arg_default_rlimit },
693 { "Manager", "DefaultLimitRSS", config_parse_rlimit, RLIMIT_RSS, arg_default_rlimit },
694 { "Manager", "DefaultLimitNOFILE", config_parse_rlimit, RLIMIT_NOFILE, arg_default_rlimit },
695 { "Manager", "DefaultLimitAS", config_parse_rlimit, RLIMIT_AS, arg_default_rlimit },
696 { "Manager", "DefaultLimitNPROC", config_parse_rlimit, RLIMIT_NPROC, arg_default_rlimit },
697 { "Manager", "DefaultLimitMEMLOCK", config_parse_rlimit, RLIMIT_MEMLOCK, arg_default_rlimit },
698 { "Manager", "DefaultLimitLOCKS", config_parse_rlimit, RLIMIT_LOCKS, arg_default_rlimit },
699 { "Manager", "DefaultLimitSIGPENDING", config_parse_rlimit, RLIMIT_SIGPENDING, arg_default_rlimit },
700 { "Manager", "DefaultLimitMSGQUEUE", config_parse_rlimit, RLIMIT_MSGQUEUE, arg_default_rlimit },
701 { "Manager", "DefaultLimitNICE", config_parse_rlimit, RLIMIT_NICE, arg_default_rlimit },
702 { "Manager", "DefaultLimitRTPRIO", config_parse_rlimit, RLIMIT_RTPRIO, arg_default_rlimit },
703 { "Manager", "DefaultLimitRTTIME", config_parse_rlimit, RLIMIT_RTTIME, arg_default_rlimit },
085afe36 704 { "Manager", "DefaultCPUAccounting", config_parse_bool, 0, &arg_default_cpu_accounting },
13c31542 705 { "Manager", "DefaultIOAccounting", config_parse_bool, 0, &arg_default_io_accounting },
377bfd2d 706 { "Manager", "DefaultIPAccounting", config_parse_bool, 0, &arg_default_ip_accounting },
085afe36
LP
707 { "Manager", "DefaultBlockIOAccounting", config_parse_bool, 0, &arg_default_blockio_accounting },
708 { "Manager", "DefaultMemoryAccounting", config_parse_bool, 0, &arg_default_memory_accounting },
03a7b521 709 { "Manager", "DefaultTasksAccounting", config_parse_bool, 0, &arg_default_tasks_accounting },
0af20ea2 710 { "Manager", "DefaultTasksMax", config_parse_tasks_max, 0, &arg_default_tasks_max },
ae8c7939 711 { "Manager", "CtrlAltDelBurstAction", config_parse_emergency_action, 0, &arg_cad_burst_action },
d3b1c508 712 {}
487393e9
LP
713 };
714
1b907b5c 715 const char *fn, *conf_dirs_nulstr;
487393e9 716
463d0d15 717 fn = arg_system ?
75eb6154
LP
718 PKGSYSCONFDIR "/system.conf" :
719 PKGSYSCONFDIR "/user.conf";
720
463d0d15 721 conf_dirs_nulstr = arg_system ?
75eb6154
LP
722 CONF_PATHS_NULSTR("systemd/system.conf.d") :
723 CONF_PATHS_NULSTR("systemd/user.conf.d");
724
bcde742e 725 (void) config_parse_many_nulstr(fn, conf_dirs_nulstr, "Manager\0", config_item_table_lookup, items, CONFIG_PARSE_WARN, NULL);
36c16a7c
LP
726
727 /* Traditionally "0" was used to turn off the default unit timeouts. Fix this up so that we used USEC_INFINITY
728 * like everywhere else. */
729 if (arg_default_timeout_start_usec <= 0)
730 arg_default_timeout_start_usec = USEC_INFINITY;
731 if (arg_default_timeout_stop_usec <= 0)
732 arg_default_timeout_stop_usec = USEC_INFINITY;
487393e9 733
487393e9
LP
734 return 0;
735}
736
85cb4151 737static void set_manager_defaults(Manager *m) {
06af2a04
TB
738
739 assert(m);
740
5b65ae15
LP
741 /* Propagates the various default unit property settings into the manager object, i.e. properties that do not
742 * affect the manager itself, but are just what newly allocated units will have set if they haven't set
743 * anything else. (Also see set_manager_settings() for the settings that affect the manager's own behaviour) */
744
06af2a04
TB
745 m->default_timer_accuracy_usec = arg_default_timer_accuracy_usec;
746 m->default_std_output = arg_default_std_output;
747 m->default_std_error = arg_default_std_error;
748 m->default_timeout_start_usec = arg_default_timeout_start_usec;
749 m->default_timeout_stop_usec = arg_default_timeout_stop_usec;
750 m->default_restart_usec = arg_default_restart_usec;
751 m->default_start_limit_interval = arg_default_start_limit_interval;
752 m->default_start_limit_burst = arg_default_start_limit_burst;
753 m->default_cpu_accounting = arg_default_cpu_accounting;
13c31542 754 m->default_io_accounting = arg_default_io_accounting;
377bfd2d 755 m->default_ip_accounting = arg_default_ip_accounting;
06af2a04
TB
756 m->default_blockio_accounting = arg_default_blockio_accounting;
757 m->default_memory_accounting = arg_default_memory_accounting;
03a7b521 758 m->default_tasks_accounting = arg_default_tasks_accounting;
0af20ea2 759 m->default_tasks_max = arg_default_tasks_max;
06af2a04
TB
760
761 manager_set_default_rlimits(m, arg_default_rlimit);
762 manager_environment_add(m, NULL, arg_default_environment);
763}
764
7b46fc6a
LP
765static void set_manager_settings(Manager *m) {
766
767 assert(m);
768
5b65ae15
LP
769 /* Propagates the various manager settings into the manager object, i.e. properties that effect the manager
770 * itself (as opposed to just being inherited into newly allocated units, see set_manager_defaults() above). */
771
7b46fc6a 772 m->confirm_spawn = arg_confirm_spawn;
2a12e32e 773 m->service_watchdogs = arg_service_watchdogs;
7b46fc6a
LP
774 m->runtime_watchdog = arg_runtime_watchdog;
775 m->shutdown_watchdog = arg_shutdown_watchdog;
776 m->cad_burst_action = arg_cad_burst_action;
777
778 manager_set_show_status(m, arg_show_status);
779}
780
f170852a 781static int parse_argv(int argc, char *argv[]) {
f170852a
LP
782 enum {
783 ARG_LOG_LEVEL = 0x100,
784 ARG_LOG_TARGET,
bbe63281
LP
785 ARG_LOG_COLOR,
786 ARG_LOG_LOCATION,
2f198e2f 787 ARG_UNIT,
edb9aaa8 788 ARG_SYSTEM,
af2d49f7 789 ARG_USER,
e537352b 790 ARG_TEST,
b87c2aa6 791 ARG_NO_PAGER,
9ba0bc4e 792 ARG_VERSION,
80876c20 793 ARG_DUMP_CONFIGURATION_ITEMS,
bbc1acab 794 ARG_DUMP_BUS_PROPERTIES,
9e58ff9c 795 ARG_DUMP_CORE,
b9e74c39 796 ARG_CRASH_CHVT,
9e58ff9c 797 ARG_CRASH_SHELL,
b9e74c39 798 ARG_CRASH_REBOOT,
a16e1123 799 ARG_CONFIRM_SPAWN,
9e58ff9c 800 ARG_SHOW_STATUS,
4288f619 801 ARG_DESERIALIZE,
2660882b 802 ARG_SWITCHED_ROOT,
0a494f1f 803 ARG_DEFAULT_STD_OUTPUT,
ee48dbd5 804 ARG_DEFAULT_STD_ERROR,
2a12e32e
JK
805 ARG_MACHINE_ID,
806 ARG_SERVICE_WATCHDOGS,
f170852a
LP
807 };
808
809 static const struct option options[] = {
a16e1123
LP
810 { "log-level", required_argument, NULL, ARG_LOG_LEVEL },
811 { "log-target", required_argument, NULL, ARG_LOG_TARGET },
bbe63281
LP
812 { "log-color", optional_argument, NULL, ARG_LOG_COLOR },
813 { "log-location", optional_argument, NULL, ARG_LOG_LOCATION },
2f198e2f 814 { "unit", required_argument, NULL, ARG_UNIT },
edb9aaa8 815 { "system", no_argument, NULL, ARG_SYSTEM },
af2d49f7 816 { "user", no_argument, NULL, ARG_USER },
a16e1123 817 { "test", no_argument, NULL, ARG_TEST },
b87c2aa6 818 { "no-pager", no_argument, NULL, ARG_NO_PAGER },
a16e1123 819 { "help", no_argument, NULL, 'h' },
9ba0bc4e 820 { "version", no_argument, NULL, ARG_VERSION },
a16e1123 821 { "dump-configuration-items", no_argument, NULL, ARG_DUMP_CONFIGURATION_ITEMS },
bbc1acab 822 { "dump-bus-properties", no_argument, NULL, ARG_DUMP_BUS_PROPERTIES },
a5d87bf0 823 { "dump-core", optional_argument, NULL, ARG_DUMP_CORE },
b9e74c39 824 { "crash-chvt", required_argument, NULL, ARG_CRASH_CHVT },
a5d87bf0 825 { "crash-shell", optional_argument, NULL, ARG_CRASH_SHELL },
b9e74c39 826 { "crash-reboot", optional_argument, NULL, ARG_CRASH_REBOOT },
a5d87bf0 827 { "confirm-spawn", optional_argument, NULL, ARG_CONFIRM_SPAWN },
6e98720f 828 { "show-status", optional_argument, NULL, ARG_SHOW_STATUS },
a16e1123 829 { "deserialize", required_argument, NULL, ARG_DESERIALIZE },
2660882b 830 { "switched-root", no_argument, NULL, ARG_SWITCHED_ROOT },
0a494f1f
LP
831 { "default-standard-output", required_argument, NULL, ARG_DEFAULT_STD_OUTPUT, },
832 { "default-standard-error", required_argument, NULL, ARG_DEFAULT_STD_ERROR, },
ee48dbd5 833 { "machine-id", required_argument, NULL, ARG_MACHINE_ID },
2a12e32e 834 { "service-watchdogs", required_argument, NULL, ARG_SERVICE_WATCHDOGS },
fb472900 835 {}
f170852a
LP
836 };
837
838 int c, r;
839
840 assert(argc >= 1);
841 assert(argv);
842
df0ff127 843 if (getpid_cached() == 1)
b770165a
LP
844 opterr = 0;
845
099663ff 846 while ((c = getopt_long(argc, argv, "hDbsz:", options, NULL)) >= 0)
f170852a
LP
847
848 switch (c) {
849
850 case ARG_LOG_LEVEL:
fb472900 851 r = log_set_max_level_from_string(optarg);
2b5107e1
ZJS
852 if (r < 0)
853 return log_error_errno(r, "Failed to parse log level \"%s\": %m", optarg);
f170852a
LP
854
855 break;
856
857 case ARG_LOG_TARGET:
fb472900 858 r = log_set_target_from_string(optarg);
2b5107e1
ZJS
859 if (r < 0)
860 return log_error_errno(r, "Failed to parse log target \"%s\": %m", optarg);
f170852a
LP
861
862 break;
863
bbe63281
LP
864 case ARG_LOG_COLOR:
865
d0b170c8 866 if (optarg) {
fb472900 867 r = log_show_color_from_string(optarg);
2b5107e1
ZJS
868 if (r < 0)
869 return log_error_errno(r, "Failed to parse log color setting \"%s\": %m",
870 optarg);
d0b170c8
LP
871 } else
872 log_show_color(true);
bbe63281
LP
873
874 break;
875
876 case ARG_LOG_LOCATION:
d0b170c8 877 if (optarg) {
fb472900 878 r = log_show_location_from_string(optarg);
2b5107e1
ZJS
879 if (r < 0)
880 return log_error_errno(r, "Failed to parse log location setting \"%s\": %m",
881 optarg);
d0b170c8
LP
882 } else
883 log_show_location(true);
bbe63281
LP
884
885 break;
886
0a494f1f 887 case ARG_DEFAULT_STD_OUTPUT:
fb472900 888 r = exec_output_from_string(optarg);
2b5107e1
ZJS
889 if (r < 0)
890 return log_error_errno(r, "Failed to parse default standard output setting \"%s\": %m",
891 optarg);
892 arg_default_std_output = r;
0a494f1f
LP
893 break;
894
895 case ARG_DEFAULT_STD_ERROR:
fb472900 896 r = exec_output_from_string(optarg);
2b5107e1
ZJS
897 if (r < 0)
898 return log_error_errno(r, "Failed to parse default standard error output setting \"%s\": %m",
899 optarg);
900 arg_default_std_error = r;
0a494f1f
LP
901 break;
902
2f198e2f 903 case ARG_UNIT:
e6e242ad 904 r = free_and_strdup(&arg_default_unit, optarg);
23bbb0de 905 if (r < 0)
2b5107e1 906 return log_error_errno(r, "Failed to set default unit \"%s\": %m", optarg);
f170852a
LP
907
908 break;
909
edb9aaa8 910 case ARG_SYSTEM:
463d0d15 911 arg_system = true;
edb9aaa8 912 break;
a5dab5ce 913
af2d49f7 914 case ARG_USER:
463d0d15 915 arg_system = false;
a5dab5ce 916 break;
a5dab5ce 917
e965d56d 918 case ARG_TEST:
fa0f4d8a 919 arg_action = ACTION_TEST;
b87c2aa6
ZJS
920 break;
921
922 case ARG_NO_PAGER:
923 arg_no_pager = true;
e965d56d
LP
924 break;
925
9ba0bc4e
ZJS
926 case ARG_VERSION:
927 arg_action = ACTION_VERSION;
928 break;
929
e537352b 930 case ARG_DUMP_CONFIGURATION_ITEMS:
fa0f4d8a 931 arg_action = ACTION_DUMP_CONFIGURATION_ITEMS;
e537352b
LP
932 break;
933
bbc1acab
YW
934 case ARG_DUMP_BUS_PROPERTIES:
935 arg_action = ACTION_DUMP_BUS_PROPERTIES;
936 break;
937
9e58ff9c 938 case ARG_DUMP_CORE:
b9e74c39
LP
939 if (!optarg)
940 arg_dump_core = true;
941 else {
942 r = parse_boolean(optarg);
943 if (r < 0)
2b5107e1
ZJS
944 return log_error_errno(r, "Failed to parse dump core boolean: \"%s\": %m",
945 optarg);
b9e74c39 946 arg_dump_core = r;
a5d87bf0 947 }
b9e74c39
LP
948 break;
949
950 case ARG_CRASH_CHVT:
951 r = parse_crash_chvt(optarg);
952 if (r < 0)
2b5107e1
ZJS
953 return log_error_errno(r, "Failed to parse crash virtual terminal index: \"%s\": %m",
954 optarg);
9e58ff9c
LP
955 break;
956
957 case ARG_CRASH_SHELL:
b9e74c39
LP
958 if (!optarg)
959 arg_crash_shell = true;
960 else {
961 r = parse_boolean(optarg);
962 if (r < 0)
2b5107e1
ZJS
963 return log_error_errno(r, "Failed to parse crash shell boolean: \"%s\": %m",
964 optarg);
b9e74c39
LP
965 arg_crash_shell = r;
966 }
967 break;
968
969 case ARG_CRASH_REBOOT:
970 if (!optarg)
971 arg_crash_reboot = true;
972 else {
973 r = parse_boolean(optarg);
974 if (r < 0)
2b5107e1
ZJS
975 return log_error_errno(r, "Failed to parse crash shell boolean: \"%s\": %m",
976 optarg);
b9e74c39 977 arg_crash_reboot = r;
a5d87bf0 978 }
9e58ff9c
LP
979 break;
980
80876c20 981 case ARG_CONFIRM_SPAWN:
7d5ceb64
FB
982 arg_confirm_spawn = mfree(arg_confirm_spawn);
983
984 r = parse_confirm_spawn(optarg, &arg_confirm_spawn);
985 if (r < 0)
2b5107e1
ZJS
986 return log_error_errno(r, "Failed to parse confirm spawn option: \"%s\": %m",
987 optarg);
80876c20
LP
988 break;
989
2a12e32e
JK
990 case ARG_SERVICE_WATCHDOGS:
991 r = parse_boolean(optarg);
992 if (r < 0)
2b5107e1
ZJS
993 return log_error_errno(r, "Failed to parse service watchdogs boolean: \"%s\": %m",
994 optarg);
2a12e32e
JK
995 arg_service_watchdogs = r;
996 break;
997
9e58ff9c 998 case ARG_SHOW_STATUS:
d450b6f2
ZJS
999 if (optarg) {
1000 r = parse_show_status(optarg, &arg_show_status);
ac7ec288 1001 if (r < 0)
2b5107e1
ZJS
1002 return log_error_errno(r, "Failed to parse show status boolean: \"%s\": %m",
1003 optarg);
d450b6f2
ZJS
1004 } else
1005 arg_show_status = SHOW_STATUS_YES;
6e98720f 1006 break;
a5d87bf0 1007
a16e1123
LP
1008 case ARG_DESERIALIZE: {
1009 int fd;
1010 FILE *f;
1011
01e10de3 1012 r = safe_atoi(optarg, &fd);
2b5107e1
ZJS
1013 if (r < 0)
1014 log_error_errno(r, "Failed to parse deserialize option \"%s\": %m", optarg);
1015 if (fd < 0) {
1016 log_error("Invalid deserialize fd: %d", fd);
b9e74c39 1017 return -EINVAL;
a16e1123
LP
1018 }
1019
b9e74c39 1020 (void) fd_cloexec(fd, true);
01e10de3
LP
1021
1022 f = fdopen(fd, "r");
4a62c710 1023 if (!f)
2b5107e1 1024 return log_error_errno(errno, "Failed to open serialization fd %d: %m", fd);
a16e1123 1025
74ca738f 1026 safe_fclose(arg_serialization);
d3b1c508 1027 arg_serialization = f;
a16e1123
LP
1028
1029 break;
1030 }
1031
2660882b 1032 case ARG_SWITCHED_ROOT:
bf4df7c3 1033 arg_switched_root = true;
d03bc1b8
HH
1034 break;
1035
ee48dbd5
NC
1036 case ARG_MACHINE_ID:
1037 r = set_machine_id(optarg);
54500613 1038 if (r < 0)
2b5107e1 1039 return log_error_errno(r, "MachineID '%s' is not valid: %m", optarg);
ee48dbd5
NC
1040 break;
1041
f170852a 1042 case 'h':
fa0f4d8a 1043 arg_action = ACTION_HELP;
f170852a
LP
1044 break;
1045
1d2e23ab
LP
1046 case 'D':
1047 log_set_max_level(LOG_DEBUG);
1048 break;
1049
099663ff
LP
1050 case 'b':
1051 case 's':
1052 case 'z':
1053 /* Just to eat away the sysvinit kernel
1054 * cmdline args without getopt() error
1055 * messages that we'll parse in
1056 * parse_proc_cmdline_word() or ignore. */
f170852a 1057
099663ff 1058 case '?':
df0ff127 1059 if (getpid_cached() != 1)
099663ff 1060 return -EINVAL;
601185b4
ZJS
1061 else
1062 return 0;
099663ff 1063
601185b4
ZJS
1064 default:
1065 assert_not_reached("Unhandled option code.");
f170852a
LP
1066 }
1067
df0ff127 1068 if (optind < argc && getpid_cached() != 1) {
d821e6d6
LP
1069 /* Hmm, when we aren't run as init system
1070 * let's complain about excess arguments */
1071
1072 log_error("Excess arguments.");
1073 return -EINVAL;
1074 }
1075
f170852a
LP
1076 return 0;
1077}
1078
1079static int help(void) {
37ec0fdd
LP
1080 _cleanup_free_ char *link = NULL;
1081 int r;
1082
1083 r = terminal_urlify_man("systemd", "1", &link);
1084 if (r < 0)
1085 return log_oom();
f170852a 1086
2e33c433 1087 printf("%s [OPTIONS...]\n\n"
af2d49f7 1088 "Starts up and maintains the system or user services.\n\n"
e537352b 1089 " -h --help Show this help\n"
cb4069d9 1090 " --version Show version\n"
e537352b 1091 " --test Determine startup sequence, dump it and exit\n"
b87c2aa6 1092 " --no-pager Do not pipe output into a pager\n"
80876c20 1093 " --dump-configuration-items Dump understood unit configuration items\n"
bbc1acab 1094 " --dump-bus-properties Dump exposed bus properties\n"
9e58ff9c 1095 " --unit=UNIT Set default unit\n"
edb9aaa8 1096 " --system Run a system instance, even if PID != 1\n"
af2d49f7 1097 " --user Run a user instance\n"
b9e74c39
LP
1098 " --dump-core[=BOOL] Dump core on crash\n"
1099 " --crash-vt=NR Change to specified VT on crash\n"
1100 " --crash-reboot[=BOOL] Reboot on crash\n"
1101 " --crash-shell[=BOOL] Run shell on crash\n"
1102 " --confirm-spawn[=BOOL] Ask for confirmation when spawning processes\n"
1103 " --show-status[=BOOL] Show status updates on the console during bootup\n"
c1dc6153 1104 " --log-target=TARGET Set log target (console, journal, kmsg, journal-or-kmsg, null)\n"
9e58ff9c 1105 " --log-level=LEVEL Set log level (debug, info, notice, warning, err, crit, alert, emerg)\n"
b9e74c39
LP
1106 " --log-color[=BOOL] Highlight important log messages\n"
1107 " --log-location[=BOOL] Include code location in log messages\n"
0a494f1f 1108 " --default-standard-output= Set default standard output for services\n"
37ec0fdd
LP
1109 " --default-standard-error= Set default standard error output for services\n"
1110 "\nSee the %s for details.\n"
1111 , program_invocation_short_name
1112 , link
1113 );
f170852a
LP
1114
1115 return 0;
1116}
1117
2cc856ac
LP
1118static int prepare_reexecute(
1119 Manager *m,
1120 FILE **ret_f,
1121 FDSet **ret_fds,
1122 bool switching_root) {
1123
48b90859
LP
1124 _cleanup_fdset_free_ FDSet *fds = NULL;
1125 _cleanup_fclose_ FILE *f = NULL;
a16e1123
LP
1126 int r;
1127
1128 assert(m);
2cc856ac
LP
1129 assert(ret_f);
1130 assert(ret_fds);
a16e1123 1131
6b78f9b4 1132 r = manager_open_serialization(m, &f);
48b90859
LP
1133 if (r < 0)
1134 return log_error_errno(r, "Failed to create serialization file: %m");
a16e1123 1135
71445ae7 1136 /* Make sure nothing is really destructed when we shut down */
313cefa1 1137 m->n_reloading++;
718db961 1138 bus_manager_send_reloading(m, true);
71445ae7 1139
6b78f9b4 1140 fds = fdset_new();
48b90859
LP
1141 if (!fds)
1142 return log_oom();
a16e1123 1143
b3680f49 1144 r = manager_serialize(m, f, fds, switching_root);
48b90859
LP
1145 if (r < 0)
1146 return log_error_errno(r, "Failed to serialize state: %m");
a16e1123 1147
48b90859
LP
1148 if (fseeko(f, 0, SEEK_SET) == (off_t) -1)
1149 return log_error_errno(errno, "Failed to rewind serialization fd: %m");
a16e1123 1150
6b78f9b4 1151 r = fd_cloexec(fileno(f), false);
48b90859
LP
1152 if (r < 0)
1153 return log_error_errno(r, "Failed to disable O_CLOEXEC for serialization: %m");
a16e1123 1154
6b78f9b4 1155 r = fdset_cloexec(fds, false);
48b90859
LP
1156 if (r < 0)
1157 return log_error_errno(r, "Failed to disable O_CLOEXEC for serialization fds: %m");
a16e1123 1158
2cc856ac
LP
1159 *ret_f = TAKE_PTR(f);
1160 *ret_fds = TAKE_PTR(fds);
a16e1123 1161
48b90859 1162 return 0;
a16e1123
LP
1163}
1164
4096d6f5 1165static int bump_rlimit_nofile(struct rlimit *saved_rlimit) {
9264cc39 1166 int r, nr;
4096d6f5
LP
1167
1168 assert(saved_rlimit);
1169
1170 /* Save the original RLIMIT_NOFILE so that we can reset it
1171 * later when transitioning from the initrd to the main
1172 * systemd or suchlike. */
4a62c710 1173 if (getrlimit(RLIMIT_NOFILE, saved_rlimit) < 0)
3ce40911 1174 return log_warning_errno(errno, "Reading RLIMIT_NOFILE failed, ignoring: %m");
4096d6f5
LP
1175
1176 /* Make sure forked processes get the default kernel setting */
1177 if (!arg_default_rlimit[RLIMIT_NOFILE]) {
1178 struct rlimit *rl;
1179
1180 rl = newdup(struct rlimit, saved_rlimit, 1);
1181 if (!rl)
1182 return log_oom();
1183
1184 arg_default_rlimit[RLIMIT_NOFILE] = rl;
1185 }
1186
9264cc39
LP
1187 /* Bump up the resource limit for ourselves substantially, all the way to the maximum the kernel allows */
1188 nr = read_nr_open();
1189 r = setrlimit_closest(RLIMIT_NOFILE, &RLIMIT_MAKE_CONST(nr));
23bbb0de 1190 if (r < 0)
3ce40911 1191 return log_warning_errno(r, "Setting RLIMIT_NOFILE failed, ignoring: %m");
4096d6f5
LP
1192
1193 return 0;
1194}
1195
fb3ae275
LP
1196static int bump_rlimit_memlock(struct rlimit *saved_rlimit) {
1197 int r;
1198
1199 assert(saved_rlimit);
1200 assert(getuid() == 0);
1201
1202 /* BPF_MAP_TYPE_LPM_TRIE bpf maps are charged against RLIMIT_MEMLOCK, even though we have CAP_IPC_LOCK which
1203 * should normally disable such checks. We need them to implement IPAccessAllow= and IPAccessDeny=, hence let's
1204 * bump the value high enough for the root user. */
1205
1206 if (getrlimit(RLIMIT_MEMLOCK, saved_rlimit) < 0)
1207 return log_warning_errno(errno, "Reading RLIMIT_MEMLOCK failed, ignoring: %m");
1208
91cfdd8d 1209 r = setrlimit_closest(RLIMIT_MEMLOCK, &RLIMIT_MAKE_CONST(1024ULL*1024ULL*64ULL));
fb3ae275
LP
1210 if (r < 0)
1211 return log_warning_errno(r, "Setting RLIMIT_MEMLOCK failed, ignoring: %m");
1212
1213 return 0;
1214}
1215
80758717 1216static void test_usr(void) {
80758717 1217
ed1c99fc 1218 /* Check that /usr is not a separate fs */
80758717 1219
871c44a7
LP
1220 if (dir_is_empty("/usr") <= 0)
1221 return;
1222
8b173b5e 1223 log_warning("/usr appears to be on its own filesystem and is not already mounted. This is not a supported setup. "
871c44a7
LP
1224 "Some things will probably break (sometimes even silently) in mysterious ways. "
1225 "Consult http://freedesktop.org/wiki/Software/systemd/separate-usr-is-broken for more information.");
1226}
1227
d3b1c508 1228static int enforce_syscall_archs(Set *archs) {
349cc4a5 1229#if HAVE_SECCOMP
d3b1c508
LP
1230 int r;
1231
83f12b27
FS
1232 if (!is_seccomp_available())
1233 return 0;
1234
469830d1 1235 r = seccomp_restrict_archs(arg_syscall_archs);
d3b1c508 1236 if (r < 0)
469830d1 1237 return log_error_errno(r, "Failed to enforce system call architecture restrication: %m");
d3b1c508 1238#endif
469830d1 1239 return 0;
d3b1c508
LP
1240}
1241
b6e2f329
LP
1242static int status_welcome(void) {
1243 _cleanup_free_ char *pretty_name = NULL, *ansi_color = NULL;
1244 int r;
1245
7a293242 1246 if (IN_SET(arg_show_status, SHOW_STATUS_NO, SHOW_STATUS_AUTO))
fd8c85c6
LP
1247 return 0;
1248
d58ad743
LP
1249 r = parse_os_release(NULL,
1250 "PRETTY_NAME", &pretty_name,
1251 "ANSI_COLOR", &ansi_color,
1252 NULL);
1253 if (r < 0)
1254 log_full_errno(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, r,
1255 "Failed to read os-release file, ignoring: %m");
b6e2f329 1256
dc9b5816
ZJS
1257 if (log_get_show_color())
1258 return status_printf(NULL, false, false,
1259 "\nWelcome to \x1B[%sm%s\x1B[0m!\n",
1260 isempty(ansi_color) ? "1" : ansi_color,
1261 isempty(pretty_name) ? "Linux" : pretty_name);
1262 else
1263 return status_printf(NULL, false, false,
1264 "\nWelcome to %s!\n",
1265 isempty(pretty_name) ? "Linux" : pretty_name);
b6e2f329
LP
1266}
1267
fdd25311
LP
1268static int write_container_id(void) {
1269 const char *c;
19854865 1270 int r;
fdd25311
LP
1271
1272 c = getenv("container");
1273 if (isempty(c))
1274 return 0;
1275
8612da97
LP
1276 RUN_WITH_UMASK(0022)
1277 r = write_string_file("/run/systemd/container", c, WRITE_STRING_FILE_CREATE);
19854865 1278 if (r < 0)
f1f849b0 1279 return log_warning_errno(r, "Failed to write /run/systemd/container, ignoring: %m");
19854865
LP
1280
1281 return 1;
1282}
1283
1284static int bump_unix_max_dgram_qlen(void) {
1285 _cleanup_free_ char *qlen = NULL;
1286 unsigned long v;
1287 int r;
1288
3130fca5
LP
1289 /* Let's bump the net.unix.max_dgram_qlen sysctl. The kernel default of 16 is simply too low. We set the value
1290 * really really early during boot, so that it is actually applied to all our sockets, including the
1291 * $NOTIFY_SOCKET one. */
19854865
LP
1292
1293 r = read_one_line_file("/proc/sys/net/unix/max_dgram_qlen", &qlen);
1294 if (r < 0)
1295 return log_warning_errno(r, "Failed to read AF_UNIX datagram queue length, ignoring: %m");
1296
1297 r = safe_atolu(qlen, &v);
1298 if (r < 0)
3130fca5 1299 return log_warning_errno(r, "Failed to parse AF_UNIX datagram queue length '%s', ignoring: %m", qlen);
19854865
LP
1300
1301 if (v >= DEFAULT_UNIX_MAX_DGRAM_QLEN)
1302 return 0;
1303
3130fca5 1304 r = write_string_filef("/proc/sys/net/unix/max_dgram_qlen", 0, "%lu", DEFAULT_UNIX_MAX_DGRAM_QLEN);
19854865
LP
1305 if (r < 0)
1306 return log_full_errno(IN_SET(r, -EROFS, -EPERM, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
1307 "Failed to bump AF_UNIX datagram queue length, ignoring: %m");
1308
1309 return 1;
fdd25311
LP
1310}
1311
32391275
FB
1312static int fixup_environment(void) {
1313 _cleanup_free_ char *term = NULL;
4dc63c4b 1314 const char *t;
32391275
FB
1315 int r;
1316
43db615b
LP
1317 /* Only fix up the environment when we are started as PID 1 */
1318 if (getpid_cached() != 1)
1319 return 0;
1320
1321 /* We expect the environment to be set correctly if run inside a container. */
84af7821
LP
1322 if (detect_container() > 0)
1323 return 0;
1324
43db615b
LP
1325 /* When started as PID1, the kernel uses /dev/console for our stdios and uses TERM=linux whatever the backend
1326 * device used by the console. We try to make a better guess here since some consoles might not have support
1327 * for color mode for example.
32391275 1328 *
43db615b 1329 * However if TERM was configured through the kernel command line then leave it alone. */
1d84ad94 1330 r = proc_cmdline_get_key("TERM", 0, &term);
32391275
FB
1331 if (r < 0)
1332 return r;
32391275 1333
4dc63c4b
LP
1334 t = term ?: default_term_for_tty("/dev/console");
1335
1336 if (setenv("TERM", t, 1) < 0)
32391275
FB
1337 return -errno;
1338
1339 return 0;
1340}
1341
6808a0bc
LP
1342static void redirect_telinit(int argc, char *argv[]) {
1343
1344 /* This is compatibility support for SysV, where calling init as a user is identical to telinit. */
1345
1346#if HAVE_SYSV_COMPAT
1347 if (getpid_cached() == 1)
1348 return;
1349
1350 if (!strstr(program_invocation_short_name, "init"))
1351 return;
1352
1353 execv(SYSTEMCTL_BINARY_PATH, argv);
1354 log_error_errno(errno, "Failed to exec " SYSTEMCTL_BINARY_PATH ": %m");
a45d7127 1355 exit(EXIT_FAILURE);
6808a0bc
LP
1356#endif
1357}
1358
4a36297c
LP
1359static int become_shutdown(
1360 const char *shutdown_verb,
7eb35049 1361 int retval) {
4a36297c
LP
1362
1363 char log_level[DECIMAL_STR_MAX(int) + 1],
e73c54b8
JK
1364 exit_code[DECIMAL_STR_MAX(uint8_t) + 1],
1365 timeout[DECIMAL_STR_MAX(usec_t) + 1];
4a36297c 1366
e73c54b8 1367 const char* command_line[13] = {
4a36297c
LP
1368 SYSTEMD_SHUTDOWN_BINARY_PATH,
1369 shutdown_verb,
e73c54b8 1370 "--timeout", timeout,
4a36297c
LP
1371 "--log-level", log_level,
1372 "--log-target",
1373 };
1374
1375 _cleanup_strv_free_ char **env_block = NULL;
e73c54b8 1376 size_t pos = 7;
4a36297c
LP
1377 int r;
1378
7eb35049 1379 assert(shutdown_verb);
234519ae 1380 assert(!command_line[pos]);
4a36297c
LP
1381 env_block = strv_copy(environ);
1382
1383 xsprintf(log_level, "%d", log_get_max_level());
e73c54b8 1384 xsprintf(timeout, "%" PRI_USEC "us", arg_default_timeout_stop_usec);
4a36297c
LP
1385
1386 switch (log_get_target()) {
1387
1388 case LOG_TARGET_KMSG:
1389 case LOG_TARGET_JOURNAL_OR_KMSG:
1390 case LOG_TARGET_SYSLOG_OR_KMSG:
1391 command_line[pos++] = "kmsg";
1392 break;
1393
1394 case LOG_TARGET_NULL:
1395 command_line[pos++] = "null";
1396 break;
1397
1398 case LOG_TARGET_CONSOLE:
1399 default:
1400 command_line[pos++] = "console";
1401 break;
1402 };
1403
1404 if (log_get_show_color())
1405 command_line[pos++] = "--log-color";
1406
1407 if (log_get_show_location())
1408 command_line[pos++] = "--log-location";
1409
1410 if (streq(shutdown_verb, "exit")) {
1411 command_line[pos++] = "--exit-code";
1412 command_line[pos++] = exit_code;
1413 xsprintf(exit_code, "%d", retval);
1414 }
1415
1416 assert(pos < ELEMENTSOF(command_line));
1417
7eb35049
LP
1418 if (streq(shutdown_verb, "reboot") &&
1419 arg_shutdown_watchdog > 0 &&
1420 arg_shutdown_watchdog != USEC_INFINITY) {
1421
4a36297c
LP
1422 char *e;
1423
1424 /* If we reboot let's set the shutdown
1425 * watchdog and tell the shutdown binary to
1426 * repeatedly ping it */
1427 r = watchdog_set_timeout(&arg_shutdown_watchdog);
1428 watchdog_close(r < 0);
1429
1430 /* Tell the binary how often to ping, ignore failure */
1431 if (asprintf(&e, "WATCHDOG_USEC="USEC_FMT, arg_shutdown_watchdog) > 0)
8a2c1fbf
EJ
1432 (void) strv_consume(&env_block, e);
1433
1434 if (arg_watchdog_device &&
1435 asprintf(&e, "WATCHDOG_DEVICE=%s", arg_watchdog_device) > 0)
1436 (void) strv_consume(&env_block, e);
4a36297c
LP
1437 } else
1438 watchdog_close(true);
1439
1440 /* Avoid the creation of new processes forked by the
1441 * kernel; at this point, we will not listen to the
1442 * signals anyway */
1443 if (detect_container() <= 0)
1444 (void) cg_uninstall_release_agent(SYSTEMD_CGROUP_CONTROLLER);
1445
1446 execve(SYSTEMD_SHUTDOWN_BINARY_PATH, (char **) command_line, env_block);
1447 return -errno;
1448}
1449
e839bafd
LP
1450static void initialize_clock(void) {
1451 int r;
1452
1453 if (clock_is_localtime(NULL) > 0) {
1454 int min;
1455
1456 /*
1457 * The very first call of settimeofday() also does a time warp in the kernel.
1458 *
1459 * In the rtc-in-local time mode, we set the kernel's timezone, and rely on external tools to take care
1460 * of maintaining the RTC and do all adjustments. This matches the behavior of Windows, which leaves
1461 * the RTC alone if the registry tells that the RTC runs in UTC.
1462 */
1463 r = clock_set_timezone(&min);
1464 if (r < 0)
1465 log_error_errno(r, "Failed to apply local time delta, ignoring: %m");
1466 else
1467 log_info("RTC configured in localtime, applying delta of %i minutes to system time.", min);
1468
1469 } else if (!in_initrd()) {
1470 /*
1471 * Do a dummy very first call to seal the kernel's time warp magic.
1472 *
1473 * Do not call this from inside the initrd. The initrd might not carry /etc/adjtime with LOCAL, but the
1474 * real system could be set up that way. In such case, we need to delay the time-warp or the sealing
1475 * until we reach the real system.
1476 *
1477 * Do no set the kernel's timezone. The concept of local time cannot be supported reliably, the time
1478 * will jump or be incorrect at every daylight saving time change. All kernel local time concepts will
1479 * be treated as UTC that way.
1480 */
1481 (void) clock_reset_timewarp();
1482 }
1483
1484 r = clock_apply_epoch();
1485 if (r < 0)
1486 log_error_errno(r, "Current system time is before build time, but cannot correct: %m");
1487 else if (r > 0)
1488 log_info("System time before build time, advancing clock.");
1489}
1490
1e41242e 1491static void initialize_coredump(bool skip_setup) {
752bcb77 1492#if ENABLE_COREDUMP
1e41242e
LP
1493 if (getpid_cached() != 1)
1494 return;
1495
1496 /* Don't limit the core dump size, so that coredump handlers such as systemd-coredump (which honour the limit)
1497 * will process core dumps for system services by default. */
1498 if (setrlimit(RLIMIT_CORE, &RLIMIT_MAKE_CONST(RLIM_INFINITY)) < 0)
1499 log_warning_errno(errno, "Failed to set RLIMIT_CORE: %m");
1500
c6885f5f
FB
1501 /* But at the same time, turn off the core_pattern logic by default, so that no
1502 * coredumps are stored until the systemd-coredump tool is enabled via
1503 * sysctl. However it can be changed via the kernel command line later so core
1504 * dumps can still be generated during early startup and in initramfs. */
1e41242e 1505 if (!skip_setup)
e557b1a6 1506 disable_coredumps();
752bcb77 1507#endif
1e41242e
LP
1508}
1509
c6885f5f
FB
1510static void initialize_core_pattern(bool skip_setup) {
1511 int r;
1512
1513 if (skip_setup || !arg_early_core_pattern)
1514 return;
1515
1516 if (getpid_cached() != 1)
1517 return;
1518
1519 r = write_string_file("/proc/sys/kernel/core_pattern", arg_early_core_pattern, 0);
1520 if (r < 0)
1521 log_warning_errno(r, "Failed to write '%s' to /proc/sys/kernel/core_pattern, ignoring: %m", arg_early_core_pattern);
1522}
1523
3c7878f9
LP
1524static void do_reexecute(
1525 int argc,
1526 char *argv[],
1527 const struct rlimit *saved_rlimit_nofile,
1528 const struct rlimit *saved_rlimit_memlock,
1529 FDSet *fds,
1530 const char *switch_root_dir,
1531 const char *switch_root_init,
1532 const char **ret_error_message) {
1533
1534 unsigned i, j, args_size;
1535 const char **args;
1536 int r;
1537
1538 assert(saved_rlimit_nofile);
1539 assert(saved_rlimit_memlock);
1540 assert(ret_error_message);
1541
1542 /* Close and disarm the watchdog, so that the new instance can reinitialize it, but doesn't get rebooted while
1543 * we do that */
1544 watchdog_close(true);
1545
1546 /* Reset the RLIMIT_NOFILE to the kernel default, so that the new systemd can pass the kernel default to its
1547 * child processes */
1548
1549 if (saved_rlimit_nofile->rlim_cur > 0)
1550 (void) setrlimit(RLIMIT_NOFILE, saved_rlimit_nofile);
1551 if (saved_rlimit_memlock->rlim_cur != (rlim_t) -1)
1552 (void) setrlimit(RLIMIT_MEMLOCK, saved_rlimit_memlock);
1553
1554 if (switch_root_dir) {
1555 /* Kill all remaining processes from the initrd, but don't wait for them, so that we can handle the
1556 * SIGCHLD for them after deserializing. */
e73c54b8 1557 broadcast_signal(SIGTERM, false, true, arg_default_timeout_stop_usec);
3c7878f9
LP
1558
1559 /* And switch root with MS_MOVE, because we remove the old directory afterwards and detach it. */
1560 r = switch_root(switch_root_dir, "/mnt", true, MS_MOVE);
1561 if (r < 0)
1562 log_error_errno(r, "Failed to switch root, trying to continue: %m");
1563 }
1564
1565 args_size = MAX(6, argc+1);
1566 args = newa(const char*, args_size);
1567
1568 if (!switch_root_init) {
1569 char sfd[DECIMAL_STR_MAX(int) + 1];
1570
1571 /* First try to spawn ourselves with the right path, and with full serialization. We do this only if
1572 * the user didn't specify an explicit init to spawn. */
1573
1574 assert(arg_serialization);
1575 assert(fds);
1576
1577 xsprintf(sfd, "%i", fileno(arg_serialization));
1578
1579 i = 0;
1580 args[i++] = SYSTEMD_BINARY_PATH;
1581 if (switch_root_dir)
1582 args[i++] = "--switched-root";
1583 args[i++] = arg_system ? "--system" : "--user";
1584 args[i++] = "--deserialize";
1585 args[i++] = sfd;
1586 args[i++] = NULL;
1587
1588 assert(i <= args_size);
1589
1590 /*
1591 * We want valgrind to print its memory usage summary before reexecution. Valgrind won't do this is on
1592 * its own on exec(), but it will do it on exit(). Hence, to ensure we get a summary here, fork() off
1593 * a child, let it exit() cleanly, so that it prints the summary, and wait() for it in the parent,
1594 * before proceeding into the exec().
1595 */
1596 valgrind_summary_hack();
1597
1598 (void) execv(args[0], (char* const*) args);
1599 log_debug_errno(errno, "Failed to execute our own binary, trying fallback: %m");
1600 }
1601
1602 /* Try the fallback, if there is any, without any serialization. We pass the original argv[] and envp[]. (Well,
1603 * modulo the ordering changes due to getopt() in argv[], and some cleanups in envp[], but let's hope that
1604 * doesn't matter.) */
1605
1606 arg_serialization = safe_fclose(arg_serialization);
1607 fds = fdset_free(fds);
1608
1609 /* Reopen the console */
1610 (void) make_console_stdio();
1611
1612 for (j = 1, i = 1; j < (unsigned) argc; j++)
1613 args[i++] = argv[j];
1614 args[i++] = NULL;
1615 assert(i <= args_size);
1616
1617 /* Reenable any blocked signals, especially important if we switch from initial ramdisk to init=... */
1618 (void) reset_all_signal_handlers();
1619 (void) reset_signal_mask();
1620
1621 if (switch_root_init) {
1622 args[0] = switch_root_init;
1623 (void) execv(args[0], (char* const*) args);
1624 log_warning_errno(errno, "Failed to execute configured init, trying fallback: %m");
1625 }
1626
1627 args[0] = "/sbin/init";
1628 (void) execv(args[0], (char* const*) args);
1629 r = -errno;
1630
1631 manager_status_printf(NULL, STATUS_TYPE_EMERGENCY,
1632 ANSI_HIGHLIGHT_RED " !! " ANSI_NORMAL,
1633 "Failed to execute /sbin/init");
1634
1635 if (r == -ENOENT) {
1636 log_warning("No /sbin/init, trying fallback");
1637
1638 args[0] = "/bin/sh";
1639 args[1] = NULL;
1640 (void) execv(args[0], (char* const*) args);
1641 log_error_errno(errno, "Failed to execute /bin/sh, giving up: %m");
1642 } else
1643 log_warning_errno(r, "Failed to execute /sbin/init, giving up: %m");
1644
1645 *ret_error_message = "Failed to execute fallback shell";
1646}
1647
7eb35049
LP
1648static int invoke_main_loop(
1649 Manager *m,
1650 bool *ret_reexecute,
1651 int *ret_retval, /* Return parameters relevant for shutting down */
1652 const char **ret_shutdown_verb, /* … */
1653 FDSet **ret_fds, /* Return parameters for reexecuting */
1654 char **ret_switch_root_dir, /* … */
1655 char **ret_switch_root_init, /* … */
1656 const char **ret_error_message) {
1657
1658 int r;
1659
1660 assert(m);
1661 assert(ret_reexecute);
1662 assert(ret_retval);
1663 assert(ret_shutdown_verb);
1664 assert(ret_fds);
1665 assert(ret_switch_root_dir);
1666 assert(ret_switch_root_init);
1667 assert(ret_error_message);
1668
1669 for (;;) {
1670 r = manager_loop(m);
1671 if (r < 0) {
1672 *ret_error_message = "Failed to run main loop";
1673 return log_emergency_errno(r, "Failed to run main loop: %m");
1674 }
1675
3ca4d0b3 1676 switch ((ManagerObjective) r) {
7eb35049 1677
a6ecbf83 1678 case MANAGER_RELOAD: {
bda7d78b 1679 LogTarget saved_log_target;
a6ecbf83
FB
1680 int saved_log_level;
1681
7eb35049
LP
1682 log_info("Reloading.");
1683
3fe91079 1684 /* First, save any overridden log level/target, then parse the configuration file, which might
bda7d78b
FB
1685 * change the log level to new settings. */
1686
a6ecbf83 1687 saved_log_level = m->log_level_overridden ? log_get_max_level() : -1;
bda7d78b 1688 saved_log_target = m->log_target_overridden ? log_get_target() : _LOG_TARGET_INVALID;
a6ecbf83 1689
7eb35049
LP
1690 r = parse_config_file();
1691 if (r < 0)
1692 log_warning_errno(r, "Failed to parse config file, ignoring: %m");
1693
1694 set_manager_defaults(m);
1695
a6ecbf83
FB
1696 if (saved_log_level >= 0)
1697 manager_override_log_level(m, saved_log_level);
bda7d78b
FB
1698 if (saved_log_target >= 0)
1699 manager_override_log_target(m, saved_log_target);
a6ecbf83 1700
7eb35049
LP
1701 r = manager_reload(m);
1702 if (r < 0)
7a35fa24
LP
1703 /* Reloading failed before the point of no return. Let's continue running as if nothing happened. */
1704 m->objective = MANAGER_OK;
7eb35049
LP
1705
1706 break;
a6ecbf83 1707 }
7eb35049
LP
1708
1709 case MANAGER_REEXECUTE:
1710
1711 r = prepare_reexecute(m, &arg_serialization, ret_fds, false);
1712 if (r < 0) {
1713 *ret_error_message = "Failed to prepare for reexecution";
1714 return r;
1715 }
1716
1717 log_notice("Reexecuting.");
1718
1719 *ret_reexecute = true;
1720 *ret_retval = EXIT_SUCCESS;
1721 *ret_shutdown_verb = NULL;
1722 *ret_switch_root_dir = *ret_switch_root_init = NULL;
1723
1724 return 0;
1725
1726 case MANAGER_SWITCH_ROOT:
1727 if (!m->switch_root_init) {
1728 r = prepare_reexecute(m, &arg_serialization, ret_fds, true);
1729 if (r < 0) {
1730 *ret_error_message = "Failed to prepare for reexecution";
1731 return r;
1732 }
1733 } else
1734 *ret_fds = NULL;
1735
1736 log_notice("Switching root.");
1737
1738 *ret_reexecute = true;
1739 *ret_retval = EXIT_SUCCESS;
1740 *ret_shutdown_verb = NULL;
1741
1742 /* Steal the switch root parameters */
1743 *ret_switch_root_dir = m->switch_root;
1744 *ret_switch_root_init = m->switch_root_init;
1745 m->switch_root = m->switch_root_init = NULL;
1746
1747 return 0;
1748
1749 case MANAGER_EXIT:
1750
1751 if (MANAGER_IS_USER(m)) {
1752 log_debug("Exit.");
1753
1754 *ret_reexecute = false;
1755 *ret_retval = m->return_value;
1756 *ret_shutdown_verb = NULL;
1757 *ret_fds = NULL;
1758 *ret_switch_root_dir = *ret_switch_root_init = NULL;
1759
1760 return 0;
1761 }
1762
1763 _fallthrough_;
1764 case MANAGER_REBOOT:
1765 case MANAGER_POWEROFF:
1766 case MANAGER_HALT:
1767 case MANAGER_KEXEC: {
af41e508
LP
1768 static const char * const table[_MANAGER_OBJECTIVE_MAX] = {
1769 [MANAGER_EXIT] = "exit",
1770 [MANAGER_REBOOT] = "reboot",
7eb35049 1771 [MANAGER_POWEROFF] = "poweroff",
af41e508
LP
1772 [MANAGER_HALT] = "halt",
1773 [MANAGER_KEXEC] = "kexec",
7eb35049
LP
1774 };
1775
1776 log_notice("Shutting down.");
1777
1778 *ret_reexecute = false;
1779 *ret_retval = m->return_value;
af41e508 1780 assert_se(*ret_shutdown_verb = table[m->objective]);
7eb35049
LP
1781 *ret_fds = NULL;
1782 *ret_switch_root_dir = *ret_switch_root_init = NULL;
1783
1784 return 0;
1785 }
1786
1787 default:
af41e508 1788 assert_not_reached("Unknown or unexpected manager objective.");
7eb35049
LP
1789 }
1790 }
1791}
1792
31aef7ff
LP
1793static void log_execution_mode(bool *ret_first_boot) {
1794 assert(ret_first_boot);
1795
1796 if (arg_system) {
1797 int v;
1798
1799 log_info(PACKAGE_STRING " running in %ssystem mode. (" SYSTEMD_FEATURES ")",
1800 arg_action == ACTION_TEST ? "test " : "" );
1801
1802 v = detect_virtualization();
1803 if (v > 0)
1804 log_info("Detected virtualization %s.", virtualization_to_string(v));
1805
1806 log_info("Detected architecture %s.", architecture_to_string(uname_architecture()));
1807
1808 if (in_initrd()) {
1809 *ret_first_boot = false;
1810 log_info("Running in initial RAM disk.");
1811 } else {
1812 /* Let's check whether we are in first boot, i.e. whether /etc is still unpopulated. We use
1813 * /etc/machine-id as flag file, for this: if it exists we assume /etc is populated, if it
1814 * doesn't it's unpopulated. This allows container managers and installers to provision a
1815 * couple of files already. If the container manager wants to provision the machine ID itself
1816 * it should pass $container_uuid to PID 1. */
1817
1818 *ret_first_boot = access("/etc/machine-id", F_OK) < 0;
1819 if (*ret_first_boot)
1820 log_info("Running with unpopulated /etc.");
1821 }
1822 } else {
b9e90f3a
LP
1823 if (DEBUG_LOGGING) {
1824 _cleanup_free_ char *t;
31aef7ff 1825
b9e90f3a
LP
1826 t = uid_to_name(getuid());
1827 log_debug(PACKAGE_STRING " running in %suser mode for user " UID_FMT "/%s. (" SYSTEMD_FEATURES ")",
1828 arg_action == ACTION_TEST ? " test" : "", getuid(), strna(t));
1829 }
31aef7ff
LP
1830
1831 *ret_first_boot = false;
1832 }
1833}
1834
5afbaa36
LP
1835static int initialize_runtime(
1836 bool skip_setup,
1837 struct rlimit *saved_rlimit_nofile,
1838 struct rlimit *saved_rlimit_memlock,
1839 const char **ret_error_message) {
1840
1841 int r;
1842
1843 assert(ret_error_message);
1844
1845 /* Sets up various runtime parameters. Many of these initializations are conditionalized:
1846 *
1847 * - Some only apply to --system instances
1848 * - Some only apply to --user instances
1849 * - Some only apply when we first start up, but not when we reexecute
1850 */
1851
2d776038
LP
1852 if (arg_action != ACTION_RUN)
1853 return 0;
1854
3c3c6cb9
LP
1855 if (arg_system) {
1856 /* Make sure we leave a core dump without panicing the kernel. */
1857 install_crash_handler();
5afbaa36 1858
3c3c6cb9
LP
1859 if (!skip_setup) {
1860 r = mount_cgroup_controllers(arg_join_controllers);
1861 if (r < 0) {
1862 *ret_error_message = "Failed to mount cgroup hierarchies";
1863 return r;
1864 }
1865
1866 status_welcome();
1867 hostname_setup();
1868 machine_id_setup(NULL, arg_machine_id, NULL);
1869 loopback_setup();
1870 bump_unix_max_dgram_qlen();
1871 test_usr();
1872 write_container_id();
1873 }
8a2c1fbf 1874
3c3c6cb9
LP
1875 if (arg_watchdog_device) {
1876 r = watchdog_set_device(arg_watchdog_device);
1877 if (r < 0)
1878 log_warning_errno(r, "Failed to set watchdog device to %s, ignoring: %m", arg_watchdog_device);
1879 }
1880
1881 if (arg_runtime_watchdog > 0 && arg_runtime_watchdog != USEC_INFINITY)
1882 watchdog_set_timeout(&arg_runtime_watchdog);
1883 }
5afbaa36
LP
1884
1885 if (arg_timer_slack_nsec != NSEC_INFINITY)
1886 if (prctl(PR_SET_TIMERSLACK, arg_timer_slack_nsec) < 0)
3a671cd1 1887 log_warning_errno(errno, "Failed to adjust timer slack, ignoring: %m");
5afbaa36
LP
1888
1889 if (arg_system && !cap_test_all(arg_capability_bounding_set)) {
1890 r = capability_bounding_set_drop_usermode(arg_capability_bounding_set);
1891 if (r < 0) {
1892 *ret_error_message = "Failed to drop capability bounding set of usermode helpers";
1893 return log_emergency_errno(r, "Failed to drop capability bounding set of usermode helpers: %m");
1894 }
1895
1896 r = capability_bounding_set_drop(arg_capability_bounding_set, true);
1897 if (r < 0) {
1898 *ret_error_message = "Failed to drop capability bounding set";
1899 return log_emergency_errno(r, "Failed to drop capability bounding set: %m");
1900 }
1901 }
1902
39362f6f
JB
1903 if (arg_system && arg_no_new_privs) {
1904 if (prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0) < 0) {
1905 *ret_error_message = "Failed to disable new privileges";
1906 return log_emergency_errno(errno, "Failed to disable new privileges: %m");
1907 }
1908 }
1909
5afbaa36
LP
1910 if (arg_syscall_archs) {
1911 r = enforce_syscall_archs(arg_syscall_archs);
1912 if (r < 0) {
1913 *ret_error_message = "Failed to set syscall architectures";
1914 return r;
1915 }
1916 }
1917
1918 if (!arg_system)
1919 /* Become reaper of our children */
1920 if (prctl(PR_SET_CHILD_SUBREAPER, 1) < 0)
1921 log_warning_errno(errno, "Failed to make us a subreaper: %m");
1922
1923 if (arg_system) {
1924 /* Bump up RLIMIT_NOFILE for systemd itself */
1925 (void) bump_rlimit_nofile(saved_rlimit_nofile);
1926 (void) bump_rlimit_memlock(saved_rlimit_memlock);
1927 }
1928
1929 return 0;
1930}
1931
6acca5fc
LP
1932static int do_queue_default_job(
1933 Manager *m,
1934 const char **ret_error_message) {
1935
1936 _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL;
1937 Job *default_unit_job;
1938 Unit *target = NULL;
1939 int r;
1940
1941 log_debug("Activating default unit: %s", arg_default_unit);
1942
4109ede7
ZJS
1943 r = manager_load_startable_unit_or_warn(m, arg_default_unit, NULL, &target);
1944 if (r < 0) {
1945 log_info("Falling back to rescue target: " SPECIAL_RESCUE_TARGET);
6acca5fc 1946
4109ede7 1947 r = manager_load_startable_unit_or_warn(m, SPECIAL_RESCUE_TARGET, NULL, &target);
6acca5fc 1948 if (r < 0) {
4109ede7
ZJS
1949 *ret_error_message = r == -ERFKILL ? "Rescue target masked"
1950 : "Failed to load rescue target";
1951 return r;
6acca5fc
LP
1952 }
1953 }
1954
1955 assert(target->load_state == UNIT_LOADED);
1956
1957 r = manager_add_job(m, JOB_START, target, JOB_ISOLATE, &error, &default_unit_job);
1958 if (r == -EPERM) {
1959 log_debug_errno(r, "Default target could not be isolated, starting instead: %s", bus_error_message(&error, r));
1960
1961 sd_bus_error_free(&error);
1962
1963 r = manager_add_job(m, JOB_START, target, JOB_REPLACE, &error, &default_unit_job);
1964 if (r < 0) {
1965 *ret_error_message = "Failed to start default target";
1966 return log_emergency_errno(r, "Failed to start default target: %s", bus_error_message(&error, r));
1967 }
1968
1969 } else if (r < 0) {
1970 *ret_error_message = "Failed to isolate default target";
1971 return log_emergency_errno(r, "Failed to isolate default target: %s", bus_error_message(&error, r));
1972 }
1973
1974 m->default_unit_job_id = default_unit_job->id;
1975
1976 return 0;
1977}
1978
970777b5 1979static void free_arguments(void) {
970777b5
LP
1980
1981 /* Frees all arg_* variables, with the exception of arg_serialization */
ef59409f 1982 rlimit_free_all(arg_default_rlimit);
970777b5
LP
1983
1984 arg_default_unit = mfree(arg_default_unit);
1985 arg_confirm_spawn = mfree(arg_confirm_spawn);
1986 arg_join_controllers = strv_free_free(arg_join_controllers);
1987 arg_default_environment = strv_free(arg_default_environment);
1988 arg_syscall_archs = set_free(arg_syscall_archs);
1989}
1990
97d1fb94
LP
1991static int load_configuration(int argc, char **argv, const char **ret_error_message) {
1992 int r;
1993
1994 assert(ret_error_message);
1995
1996 arg_default_tasks_max = system_tasks_max_scale(DEFAULT_TASKS_MAX_PERCENTAGE, 100U);
1997
1998 r = parse_config_file();
1999 if (r < 0) {
2000 *ret_error_message = "Failed to parse config file";
2001 return r;
2002 }
2003
2004 if (arg_system) {
2005 r = proc_cmdline_parse(parse_proc_cmdline_item, NULL, 0);
2006 if (r < 0)
2007 log_warning_errno(r, "Failed to parse kernel command line, ignoring: %m");
2008 }
2009
2010 /* Note that this also parses bits from the kernel command line, including "debug". */
2011 log_parse_environment();
2012
2013 r = parse_argv(argc, argv);
2014 if (r < 0) {
2015 *ret_error_message = "Failed to parse commandline arguments";
2016 return r;
2017 }
2018
2019 /* Initialize default unit */
2020 if (!arg_default_unit) {
2021 arg_default_unit = strdup(SPECIAL_DEFAULT_TARGET);
2022 if (!arg_default_unit) {
2023 *ret_error_message = "Failed to set default unit";
2024 return log_oom();
2025 }
2026 }
2027
db33214b 2028 /* Initialize the show status setting if it hasn't been set explicitly yet */
7a293242 2029 if (arg_show_status == _SHOW_STATUS_INVALID)
db33214b
LP
2030 arg_show_status = SHOW_STATUS_YES;
2031
97d1fb94
LP
2032 return 0;
2033}
2034
b0d7c989
LP
2035static int safety_checks(void) {
2036
febf46a4
LP
2037 if (getpid_cached() == 1 &&
2038 arg_action != ACTION_RUN) {
2039 log_error("Unsupported execution mode while PID 1.");
2040 return -EPERM;
2041 }
2042
2043 if (getpid_cached() == 1 &&
2044 !arg_system) {
2045 log_error("Can't run --user mode as PID 1.");
2046 return -EPERM;
2047 }
2048
2049 if (arg_action == ACTION_RUN &&
2050 arg_system &&
2051 getpid_cached() != 1) {
2052 log_error("Can't run system mode unless PID 1.");
2053 return -EPERM;
2054 }
2055
b0d7c989
LP
2056 if (arg_action == ACTION_TEST &&
2057 geteuid() == 0) {
2058 log_error("Don't run test mode as root.");
2059 return -EPERM;
2060 }
2061
2062 if (!arg_system &&
2063 arg_action == ACTION_RUN &&
2064 sd_booted() <= 0) {
2065 log_error("Trying to run as user instance, but the system has not been booted with systemd.");
2066 return -EOPNOTSUPP;
2067 }
2068
2069 if (!arg_system &&
2070 arg_action == ACTION_RUN &&
2071 !getenv("XDG_RUNTIME_DIR")) {
2072 log_error("Trying to run as user instance, but $XDG_RUNTIME_DIR is not set.");
2073 return -EUNATCH;
2074 }
2075
2076 if (arg_system &&
2077 arg_action == ACTION_RUN &&
2078 running_in_chroot() > 0) {
2079 log_error("Cannot be run in a chroot() environment.");
2080 return -EOPNOTSUPP;
2081 }
2082
2083 return 0;
2084}
2085
74da609f
LP
2086static int initialize_security(
2087 bool *loaded_policy,
2088 dual_timestamp *security_start_timestamp,
2089 dual_timestamp *security_finish_timestamp,
2090 const char **ret_error_message) {
2091
2092 int r;
2093
2094 assert(loaded_policy);
2095 assert(security_start_timestamp);
2096 assert(security_finish_timestamp);
2097 assert(ret_error_message);
2098
2099 dual_timestamp_get(security_start_timestamp);
2100
97149f40 2101 r = mac_selinux_setup(loaded_policy);
74da609f
LP
2102 if (r < 0) {
2103 *ret_error_message = "Failed to load SELinux policy";
2104 return r;
2105 }
2106
2107 r = mac_smack_setup(loaded_policy);
2108 if (r < 0) {
2109 *ret_error_message = "Failed to load SMACK policy";
2110 return r;
2111 }
2112
2113 r = ima_setup();
2114 if (r < 0) {
2115 *ret_error_message = "Failed to load IMA policy";
2116 return r;
2117 }
2118
2119 dual_timestamp_get(security_finish_timestamp);
2120 return 0;
2121}
2122
263162da
LP
2123static void test_summary(Manager *m) {
2124 assert(m);
2125
2126 printf("-> By units:\n");
2127 manager_dump_units(m, stdout, "\t");
2128
2129 printf("-> By jobs:\n");
2130 manager_dump_jobs(m, stdout, "\t");
2131}
2132
efeb853f
LP
2133static int collect_fds(FDSet **ret_fds, const char **ret_error_message) {
2134 int r;
2135
2136 assert(ret_fds);
2137 assert(ret_error_message);
2138
2139 r = fdset_new_fill(ret_fds);
2140 if (r < 0) {
2141 *ret_error_message = "Failed to allocate fd set";
2142 return log_emergency_errno(r, "Failed to allocate fd set: %m");
2143 }
2144
2145 fdset_cloexec(*ret_fds, true);
2146
2147 if (arg_serialization)
2148 assert_se(fdset_remove(*ret_fds, fileno(arg_serialization)) >= 0);
2149
2150 return 0;
2151}
2152
2e51b31c
LP
2153static void setup_console_terminal(bool skip_setup) {
2154
2155 if (!arg_system)
2156 return;
2157
2158 /* Become a session leader if we aren't one yet. */
2159 (void) setsid();
2160
2161 /* If we are init, we connect stdin/stdout/stderr to /dev/null and make sure we don't have a controlling
2162 * tty. */
2163 (void) release_terminal();
2164
2165 /* Reset the console, but only if this is really init and we are freshly booted */
2166 if (getpid_cached() == 1 && !skip_setup)
2167 (void) console_setup();
2168}
2169
aa40ff07
LP
2170static bool early_skip_setup_check(int argc, char *argv[]) {
2171 bool found_deserialize = false;
2172 int i;
2173
2174 /* Determine if this is a reexecution or normal bootup. We do the full command line parsing much later, so
2175 * let's just have a quick peek here. Note that if we have switched root, do all the special setup things
2176 * anyway, even if in that case we also do deserialization. */
2177
2178 for (i = 1; i < argc; i++) {
aa40ff07
LP
2179 if (streq(argv[i], "--switched-root"))
2180 return false; /* If we switched root, don't skip the setup. */
2181 else if (streq(argv[i], "--deserialize"))
2182 found_deserialize = true;
2183 }
2184
2185 return found_deserialize; /* When we are deserializing, then we are reexecuting, hence avoid the extensive setup */
2186}
2187
60918275 2188int main(int argc, char *argv[]) {
625e8690
LP
2189
2190 dual_timestamp initrd_timestamp = DUAL_TIMESTAMP_NULL, userspace_timestamp = DUAL_TIMESTAMP_NULL, kernel_timestamp = DUAL_TIMESTAMP_NULL,
2191 security_start_timestamp = DUAL_TIMESTAMP_NULL, security_finish_timestamp = DUAL_TIMESTAMP_NULL;
2192 struct rlimit saved_rlimit_nofile = RLIMIT_MAKE_CONST(0), saved_rlimit_memlock = RLIMIT_MAKE_CONST((rlim_t) -1);
2193 bool skip_setup, loaded_policy = false, queue_default_job = false, first_boot = false, reexecute = false;
2194 char *switch_root_dir = NULL, *switch_root_init = NULL;
9d76d730 2195 usec_t before_startup, after_startup;
625e8690 2196 static char systemd[] = "systemd";
9d76d730 2197 char timespan[FORMAT_TIMESPAN_MAX];
625e8690
LP
2198 const char *shutdown_verb = NULL, *error_message = NULL;
2199 int r, retval = EXIT_FAILURE;
2200 Manager *m = NULL;
a16e1123 2201 FDSet *fds = NULL;
27b14a22 2202
d72a8f10 2203 /* SysV compatibility: redirect init → telinit */
6808a0bc 2204 redirect_telinit(argc, argv);
2cb1a60d 2205
d72a8f10 2206 /* Take timestamps early on */
c3a170f3
HH
2207 dual_timestamp_from_monotonic(&kernel_timestamp, 0);
2208 dual_timestamp_get(&userspace_timestamp);
2209
d72a8f10
LP
2210 /* Figure out whether we need to do initialize the system, or if we already did that because we are
2211 * reexecuting */
aa40ff07 2212 skip_setup = early_skip_setup_check(argc, argv);
d03bc1b8 2213
d72a8f10
LP
2214 /* If we get started via the /sbin/init symlink then we are called 'init'. After a subsequent reexecution we
2215 * are then called 'systemd'. That is confusing, hence let's call us systemd right-away. */
f3b6a3ed 2216 program_invocation_short_name = systemd;
eee8b7ab 2217 (void) prctl(PR_SET_NAME, systemd);
5d6b1584 2218
d72a8f10 2219 /* Save the original command line */
9a0e6896
LP
2220 saved_argv = argv;
2221 saved_argc = argc;
f3b6a3ed 2222
6fdb8de4 2223 /* Make sure that if the user says "syslog" we actually log to the journal. */
c1dc6153 2224 log_set_upgrade_syslog_to_journal(true);
bbe63281 2225
df0ff127 2226 if (getpid_cached() == 1) {
b5752d23
LP
2227 /* When we run as PID 1 force system mode */
2228 arg_system = true;
2229
48a601fe 2230 /* Disable the umask logic */
90dc8c2e
MG
2231 umask(0);
2232
92890452
LP
2233 /* Make sure that at least initially we do not ever log to journald/syslogd, because it might not be
2234 * activated yet (even though the log socket for it exists). */
d075092f
LP
2235 log_set_prohibit_ipc(true);
2236
48a601fe
LP
2237 /* Always reopen /dev/console when running as PID 1 or one of its pre-execve() children. This is
2238 * important so that we never end up logging to any foreign stderr, for example if we have to log in a
2239 * child process right before execve()'ing the actual binary, at a point in time where socket
2240 * activation stderr/stdout area already set up. */
2241 log_set_always_reopen_console(true);
48a601fe 2242
92890452 2243 if (detect_container() <= 0) {
4f8d551f 2244
92890452 2245 /* Running outside of a container as PID 1 */
92890452
LP
2246 log_set_target(LOG_TARGET_KMSG);
2247 log_open();
a866073d 2248
92890452
LP
2249 if (in_initrd())
2250 initrd_timestamp = userspace_timestamp;
c3ba6250 2251
92890452
LP
2252 if (!skip_setup) {
2253 r = mount_setup_early();
2254 if (r < 0) {
2255 error_message = "Failed to mount early API filesystems";
2256 goto finish;
2257 }
2258
2259 r = initialize_security(
2260 &loaded_policy,
2261 &security_start_timestamp,
2262 &security_finish_timestamp,
2263 &error_message);
2264 if (r < 0)
2265 goto finish;
d723cd65 2266 }
eee8b7ab 2267
92890452
LP
2268 if (mac_selinux_init() < 0) {
2269 error_message = "Failed to initialize SELinux policy";
96694e99 2270 goto finish;
92890452 2271 }
0b3325e7 2272
92890452
LP
2273 if (!skip_setup)
2274 initialize_clock();
2275
2276 /* Set the default for later on, but don't actually open the logs like this for now. Note that
2277 * if we are transitioning from the initrd there might still be journal fd open, and we
2278 * shouldn't attempt opening that before we parsed /proc/cmdline which might redirect output
2279 * elsewhere. */
2280 log_set_target(LOG_TARGET_JOURNAL_OR_KMSG);
2281
2282 } else {
2283 /* Running inside a container, as PID 1 */
92890452
LP
2284 log_set_target(LOG_TARGET_CONSOLE);
2285 log_open();
2286
2287 /* For later on, see above... */
2288 log_set_target(LOG_TARGET_JOURNAL);
2289
2290 /* clear the kernel timestamp,
2291 * because we are in a container */
2292 kernel_timestamp = DUAL_TIMESTAMP_NULL;
cb6531be 2293 }
7948c4df 2294
92890452 2295 initialize_coredump(skip_setup);
a866073d 2296
92890452
LP
2297 r = fixup_environment();
2298 if (r < 0) {
2299 log_emergency_errno(r, "Failed to fix up PID 1 environment: %m");
2300 error_message = "Failed to fix up PID1 environment";
2301 goto finish;
2302 }
a866073d 2303
c3a170f3 2304 } else {
a866073d 2305 /* Running as user instance */
463d0d15 2306 arg_system = false;
eeecf6e6 2307 log_set_target(LOG_TARGET_AUTO);
871e5809 2308 log_open();
c3a170f3
HH
2309
2310 /* clear the kernel timestamp,
2311 * because we are not PID 1 */
6513d561 2312 kernel_timestamp = DUAL_TIMESTAMP_NULL;
bbe63281 2313 }
a5dab5ce 2314
6edefe0b 2315 if (arg_system) {
92890452
LP
2316 /* Try to figure out if we can use colors with the console. No need to do that for user instances since
2317 * they never log into the console. */
3a18b604 2318 log_show_color(colors_enabled());
92890452 2319
c76cf844
AK
2320 r = make_null_stdio();
2321 if (r < 0)
92890452 2322 log_warning_errno(r, "Failed to redirect standard streams to /dev/null, ignoring: %m");
3a18b604
FB
2323 }
2324
f170852a
LP
2325 /* Mount /proc, /sys and friends, so that /proc/cmdline and
2326 * /proc/$PID/fd is available. */
df0ff127 2327 if (getpid_cached() == 1) {
f84f9974 2328
a132bef0 2329 /* Load the kernel modules early. */
2e75e2a8
DM
2330 if (!skip_setup)
2331 kmod_setup();
2e75e2a8 2332
0c85a4f3 2333 r = mount_setup(loaded_policy);
cb6531be
ZJS
2334 if (r < 0) {
2335 error_message = "Failed to mount API filesystems";
8efe3c01 2336 goto finish;
cb6531be 2337 }
0c85a4f3 2338 }
4ade7963
LP
2339
2340 /* Reset all signal handlers. */
ce30c8dc
LP
2341 (void) reset_all_signal_handlers();
2342 (void) ignore_signals(SIGNALS_IGNORE, -1);
078e4539 2343
97d1fb94
LP
2344 r = load_configuration(argc, argv, &error_message);
2345 if (r < 0)
f170852a 2346 goto finish;
10c961b9 2347
b0d7c989
LP
2348 r = safety_checks();
2349 if (r < 0)
fe783b03 2350 goto finish;
fe783b03 2351
bbc1acab 2352 if (IN_SET(arg_action, ACTION_TEST, ACTION_HELP, ACTION_DUMP_CONFIGURATION_ITEMS, ACTION_DUMP_BUS_PROPERTIES))
ee5324aa 2353 (void) pager_open(arg_no_pager, false);
b0d7c989
LP
2354
2355 if (arg_action != ACTION_RUN)
74e7579c 2356 skip_setup = true;
b87c2aa6 2357
fa0f4d8a 2358 if (arg_action == ACTION_HELP) {
37ec0fdd 2359 retval = help() < 0 ? EXIT_FAILURE : EXIT_SUCCESS;
f170852a 2360 goto finish;
9ba0bc4e
ZJS
2361 } else if (arg_action == ACTION_VERSION) {
2362 retval = version();
2363 goto finish;
fa0f4d8a 2364 } else if (arg_action == ACTION_DUMP_CONFIGURATION_ITEMS) {
e537352b 2365 unit_dump_config_items(stdout);
22f4096c 2366 retval = EXIT_SUCCESS;
e537352b 2367 goto finish;
bbc1acab
YW
2368 } else if (arg_action == ACTION_DUMP_BUS_PROPERTIES) {
2369 dump_bus_properties(stdout);
2370 retval = EXIT_SUCCESS;
2371 goto finish;
f170852a
LP
2372 }
2373
4c701096 2374 assert_se(IN_SET(arg_action, ACTION_RUN, ACTION_TEST));
f170852a 2375
5a2e0c62
LP
2376 /* Move out of the way, so that we won't block unmounts */
2377 assert_se(chdir("/") == 0);
2378
dea374e8 2379 if (arg_action == ACTION_RUN) {
a70c72a0 2380
c6885f5f
FB
2381 /* A core pattern might have been specified via the cmdline. */
2382 initialize_core_pattern(skip_setup);
2383
efeb853f 2384 /* Close logging fds, in order not to confuse collecting passed fds and terminal logic below */
a70c72a0
LP
2385 log_close();
2386
2387 /* Remember open file descriptors for later deserialization */
efeb853f
LP
2388 r = collect_fds(&fds, &error_message);
2389 if (r < 0)
dea374e8 2390 goto finish;
a16e1123 2391
2e51b31c
LP
2392 /* Give up any control of the console, but make sure its initialized. */
2393 setup_console_terminal(skip_setup);
56d96fc0 2394
a70c72a0
LP
2395 /* Open the logging devices, if possible and necessary */
2396 log_open();
56d96fc0 2397 }
4ade7963 2398
31aef7ff 2399 log_execution_mode(&first_boot);
a5dab5ce 2400
2d776038
LP
2401 r = initialize_runtime(skip_setup,
2402 &saved_rlimit_nofile,
2403 &saved_rlimit_memlock,
2404 &error_message);
2405 if (r < 0)
2406 goto finish;
4096d6f5 2407
e0a3da1f
ZJS
2408 r = manager_new(arg_system ? UNIT_FILE_SYSTEM : UNIT_FILE_USER,
2409 arg_action == ACTION_TEST ? MANAGER_TEST_FULL : 0,
2410 &m);
e96d6be7 2411 if (r < 0) {
da927ba9 2412 log_emergency_errno(r, "Failed to allocate manager object: %m");
cb6531be 2413 error_message = "Failed to allocate manager object";
60918275
LP
2414 goto finish;
2415 }
2416
9f9f0342
LP
2417 m->timestamps[MANAGER_TIMESTAMP_KERNEL] = kernel_timestamp;
2418 m->timestamps[MANAGER_TIMESTAMP_INITRD] = initrd_timestamp;
2419 m->timestamps[MANAGER_TIMESTAMP_USERSPACE] = userspace_timestamp;
d4ee7bd8
YW
2420 m->timestamps[manager_timestamp_initrd_mangle(MANAGER_TIMESTAMP_SECURITY_START)] = security_start_timestamp;
2421 m->timestamps[manager_timestamp_initrd_mangle(MANAGER_TIMESTAMP_SECURITY_FINISH)] = security_finish_timestamp;
9e58ff9c 2422
85cb4151 2423 set_manager_defaults(m);
7b46fc6a 2424 set_manager_settings(m);
fd130612 2425 manager_set_first_boot(m, first_boot);
27d340c7 2426
bf4df7c3 2427 /* Remember whether we should queue the default job */
d3b1c508 2428 queue_default_job = !arg_serialization || arg_switched_root;
bf4df7c3 2429
9d76d730
LP
2430 before_startup = now(CLOCK_MONOTONIC);
2431
d3b1c508 2432 r = manager_startup(m, arg_serialization, fds);
58f88d92 2433 if (r < 0) {
cefb3eda 2434 error_message = "Failed to start up manager";
58f88d92
ZJS
2435 goto finish;
2436 }
a16e1123 2437
6acca5fc 2438 /* This will close all file descriptors that were opened, but not claimed by any unit. */
2feceb5e 2439 fds = fdset_free(fds);
74ca738f 2440 arg_serialization = safe_fclose(arg_serialization);
bf4df7c3
LP
2441
2442 if (queue_default_job) {
6acca5fc 2443 r = do_queue_default_job(m, &error_message);
718db961 2444 if (r < 0)
37d88da7 2445 goto finish;
6acca5fc 2446 }
ab17a050 2447
6acca5fc 2448 after_startup = now(CLOCK_MONOTONIC);
60918275 2449
6acca5fc
LP
2450 log_full(arg_action == ACTION_TEST ? LOG_INFO : LOG_DEBUG,
2451 "Loaded units and determined initial transaction in %s.",
2452 format_timespan(timespan, sizeof(timespan), after_startup - before_startup, 100 * USEC_PER_MSEC));
07672f49 2453
6acca5fc 2454 if (arg_action == ACTION_TEST) {
263162da 2455 test_summary(m);
6acca5fc
LP
2456 retval = EXIT_SUCCESS;
2457 goto finish;
e965d56d 2458 }
d46de8a1 2459
3046b6db
LP
2460 (void) invoke_main_loop(m,
2461 &reexecute,
2462 &retval,
2463 &shutdown_verb,
2464 &fds,
2465 &switch_root_dir,
2466 &switch_root_init,
2467 &error_message);
f170852a 2468
60918275 2469finish:
b87c2aa6
ZJS
2470 pager_close();
2471
92890452 2472 if (m) {
81f5fc2d 2473 arg_shutdown_watchdog = m->shutdown_watchdog;
92890452
LP
2474 m = manager_free(m);
2475 }
60918275 2476
970777b5 2477 free_arguments();
cc56fafe 2478 mac_selinux_finish();
b2bb3dbe 2479
3c7878f9
LP
2480 if (reexecute)
2481 do_reexecute(argc, argv,
2482 &saved_rlimit_nofile,
2483 &saved_rlimit_memlock,
2484 fds,
2485 switch_root_dir,
2486 switch_root_init,
2487 &error_message); /* This only returns if reexecution failed */
a16e1123 2488
74ca738f 2489 arg_serialization = safe_fclose(arg_serialization);
2feceb5e 2490 fds = fdset_free(fds);
a16e1123 2491
349cc4a5 2492#if HAVE_VALGRIND_VALGRIND_H
54b434b1
LP
2493 /* If we are PID 1 and running under valgrind, then let's exit
2494 * here explicitly. valgrind will only generate nice output on
2495 * exit(), not on exec(), hence let's do the former not the
2496 * latter here. */
8a2c1fbf
EJ
2497 if (getpid_cached() == 1 && RUNNING_ON_VALGRIND) {
2498 /* Cleanup watchdog_device strings for valgrind. We need them
2499 * in become_shutdown() so normally we cannot free them yet. */
2500 watchdog_free_device();
2501 arg_watchdog_device = mfree(arg_watchdog_device);
27fe58b7 2502 return retval;
8a2c1fbf 2503 }
54b434b1
LP
2504#endif
2505
b9080b03 2506 if (shutdown_verb) {
7eb35049 2507 r = become_shutdown(shutdown_verb, retval);
4a36297c 2508 log_error_errno(r, "Failed to execute shutdown binary, %s: %m", getpid_cached() == 1 ? "freezing" : "quitting");
9b9881d7 2509 error_message = "Failed to execute shutdown binary";
b9080b03
FF
2510 }
2511
8a2c1fbf
EJ
2512 watchdog_free_device();
2513 arg_watchdog_device = mfree(arg_watchdog_device);
2514
df0ff127 2515 if (getpid_cached() == 1) {
cb6531be
ZJS
2516 if (error_message)
2517 manager_status_printf(NULL, STATUS_TYPE_EMERGENCY,
1fc464f6 2518 ANSI_HIGHLIGHT_RED "!!!!!!" ANSI_NORMAL,
cb6531be 2519 "%s, freezing.", error_message);
b9e74c39 2520 freeze_or_reboot();
cb6531be 2521 }
c3b3c274 2522
60918275
LP
2523 return retval;
2524}