]> git.ipfire.org Git - thirdparty/systemd.git/blame - src/core/main.c
projects / thirdparty / systemd.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
mkosi: make kmsg work in our mkosi builds at least
[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
52d62075 1170 /* Save the original RLIMIT_NOFILE so that we can reset it later when transitioning from the initrd to the main
4096d6f5 1171 * systemd or suchlike. */
4a62c710 1172 if (getrlimit(RLIMIT_NOFILE, saved_rlimit) < 0)
3ce40911 1173 return log_warning_errno(errno, "Reading RLIMIT_NOFILE failed, ignoring: %m");
4096d6f5 1174
52d62075
LP		 1175 /* Get the underlying absolute limit the kernel enforces */
1176 nr = read_nr_open();
1177
1178 /* Make sure forked processes get limits based on the original kernel setting */
4096d6f5
LP		 1179 if (!arg_default_rlimit[RLIMIT_NOFILE]) {
1180 struct rlimit *rl;
1181
1182 rl = newdup(struct rlimit, saved_rlimit, 1);
1183 if (!rl)
1184 return log_oom();
1185
52d62075
LP		 1186 /* Bump the hard limit for system services to a substantially higher value. The default hard limit
1187 * current kernels set is pretty low (4K), mostly for historical reasons. According to kernel
1188 * developers, the fd handling in recent kernels has been optimized substantially enough, so that we
1189 * can bump the limit now, without paying too high a price in memory or performance. Note however that
1190 * we only bump the hard limit, not the soft limit. That's because select() works the way it works, and
1191 * chokes on fds >= 1024. If we'd bump the soft limit globally, it might accidentally happen to
1192 * unexpecting programs that they get fds higher than what they can process using select(). By only
1193 * bumping the hard limit but leaving the low limit as it is we avoid this pitfall: programs that are
1194 * written by folks aware of the select() problem in mind (and thus use poll()/epoll instead of
1195 * select(), the way everybody should) can explicitly opt into high fds by bumping their soft limit
1196 * beyond 1024, to the hard limit we pass. */
1197 if (arg_system)
1198 rl->rlim_max = MIN((rlim_t) nr, MAX(rl->rlim_max, (rlim_t) HIGH_RLIMIT_NOFILE));
1199
4096d6f5
LP		 1200 arg_default_rlimit[RLIMIT_NOFILE] = rl;
1201 }
1202
52d62075
LP		 1203 /* Bump up the resource limit for ourselves substantially, all the way to the maximum the kernel allows, for
1204 * both hard and soft. */
9264cc39 1205 r = setrlimit_closest(RLIMIT_NOFILE, &RLIMIT_MAKE_CONST(nr));
23bbb0de 1206 if (r < 0)
3ce40911 1207 return log_warning_errno(r, "Setting RLIMIT_NOFILE failed, ignoring: %m");
4096d6f5
LP		 1208
1209 return 0;
1210}
1211
fb3ae275
LP		 1212static int bump_rlimit_memlock(struct rlimit *saved_rlimit) {
1213 int r;
1214
1215 assert(saved_rlimit);
fb3ae275 1216
a17c1712
LP		 1217 /* BPF_MAP_TYPE_LPM_TRIE bpf maps are charged against RLIMIT_MEMLOCK, even if we have CAP_IPC_LOCK which should
1218 * normally disable such checks. We need them to implement IPAccessAllow= and IPAccessDeny=, hence let's bump
1219 * the value high enough for our user. */
fb3ae275
LP		 1220
1221 if (getrlimit(RLIMIT_MEMLOCK, saved_rlimit) < 0)
1222 return log_warning_errno(errno, "Reading RLIMIT_MEMLOCK failed, ignoring: %m");
1223
91cfdd8d 1224 r = setrlimit_closest(RLIMIT_MEMLOCK, &RLIMIT_MAKE_CONST(1024ULL*1024ULL*64ULL));
fb3ae275
LP		 1225 if (r < 0)
1226 return log_warning_errno(r, "Setting RLIMIT_MEMLOCK failed, ignoring: %m");
1227
1228 return 0;
1229}
1230
80758717 1231static void test_usr(void) {
80758717 1232
ed1c99fc 1233 /* Check that /usr is not a separate fs */
80758717 1234
871c44a7
LP		 1235 if (dir_is_empty("/usr") <= 0)
1236 return;
1237
8b173b5e 1238 log_warning("/usr appears to be on its own filesystem and is not already mounted. This is not a supported setup. "
871c44a7
LP		 1239 "Some things will probably break (sometimes even silently) in mysterious ways. "
1240 "Consult http://freedesktop.org/wiki/Software/systemd/separate-usr-is-broken for more information.");
1241}
1242
d3b1c508 1243static int enforce_syscall_archs(Set *archs) {
349cc4a5 1244#if HAVE_SECCOMP
d3b1c508
LP		 1245 int r;
1246
83f12b27
FS		 1247 if (!is_seccomp_available())
1248 return 0;
1249
469830d1 1250 r = seccomp_restrict_archs(arg_syscall_archs);
d3b1c508 1251 if (r < 0)
469830d1 1252 return log_error_errno(r, "Failed to enforce system call architecture restrication: %m");
d3b1c508 1253#endif
469830d1 1254 return 0;
d3b1c508
LP		 1255}
1256
b6e2f329
LP		 1257static int status_welcome(void) {
1258 _cleanup_free_ char *pretty_name = NULL, *ansi_color = NULL;
1259 int r;
1260
7a293242 1261 if (IN_SET(arg_show_status, SHOW_STATUS_NO, SHOW_STATUS_AUTO))
fd8c85c6
LP		 1262 return 0;
1263
d58ad743
LP		 1264 r = parse_os_release(NULL,
1265 "PRETTY_NAME", &pretty_name,
1266 "ANSI_COLOR", &ansi_color,
1267 NULL);
1268 if (r < 0)
1269 log_full_errno(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, r,
1270 "Failed to read os-release file, ignoring: %m");
b6e2f329 1271
dc9b5816
ZJS		 1272 if (log_get_show_color())
1273 return status_printf(NULL, false, false,
1274 "\nWelcome to \x1B[%sm%s\x1B[0m!\n",
1275 isempty(ansi_color) ? "1" : ansi_color,
1276 isempty(pretty_name) ? "Linux" : pretty_name);
1277 else
1278 return status_printf(NULL, false, false,
1279 "\nWelcome to %s!\n",
1280 isempty(pretty_name) ? "Linux" : pretty_name);
b6e2f329
LP		 1281}
1282
fdd25311
LP		 1283static int write_container_id(void) {
1284 const char *c;
19854865 1285 int r;
fdd25311
LP		 1286
1287 c = getenv("container");
1288 if (isempty(c))
1289 return 0;
1290
8612da97
LP		 1291 RUN_WITH_UMASK(0022)
1292 r = write_string_file("/run/systemd/container", c, WRITE_STRING_FILE_CREATE);
19854865 1293 if (r < 0)
f1f849b0 1294 return log_warning_errno(r, "Failed to write /run/systemd/container, ignoring: %m");
19854865
LP		 1295
1296 return 1;
1297}
1298
1299static int bump_unix_max_dgram_qlen(void) {
1300 _cleanup_free_ char *qlen = NULL;
1301 unsigned long v;
1302 int r;
1303
3130fca5
LP		 1304 /* Let's bump the net.unix.max_dgram_qlen sysctl. The kernel default of 16 is simply too low. We set the value
1305 * really really early during boot, so that it is actually applied to all our sockets, including the
1306 * $NOTIFY_SOCKET one. */
19854865
LP		 1307
1308 r = read_one_line_file("/proc/sys/net/unix/max_dgram_qlen", &qlen);
1309 if (r < 0)
1310 return log_warning_errno(r, "Failed to read AF_UNIX datagram queue length, ignoring: %m");
1311
1312 r = safe_atolu(qlen, &v);
1313 if (r < 0)
3130fca5 1314 return log_warning_errno(r, "Failed to parse AF_UNIX datagram queue length '%s', ignoring: %m", qlen);
19854865
LP		 1315
1316 if (v >= DEFAULT_UNIX_MAX_DGRAM_QLEN)
1317 return 0;
1318
3130fca5 1319 r = write_string_filef("/proc/sys/net/unix/max_dgram_qlen", 0, "%lu", DEFAULT_UNIX_MAX_DGRAM_QLEN);
19854865
LP		 1320 if (r < 0)
1321 return log_full_errno(IN_SET(r, -EROFS, -EPERM, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
1322 "Failed to bump AF_UNIX datagram queue length, ignoring: %m");
1323
1324 return 1;
fdd25311
LP		 1325}
1326
32391275
FB		 1327static int fixup_environment(void) {
1328 _cleanup_free_ char *term = NULL;
4dc63c4b 1329 const char *t;
32391275
FB		 1330 int r;
1331
43db615b
LP		 1332 /* Only fix up the environment when we are started as PID 1 */
1333 if (getpid_cached() != 1)
1334 return 0;
1335
1336 /* We expect the environment to be set correctly if run inside a container. */
84af7821
LP		 1337 if (detect_container() > 0)
1338 return 0;
1339
43db615b
LP		 1340 /* When started as PID1, the kernel uses /dev/console for our stdios and uses TERM=linux whatever the backend
1341 * device used by the console. We try to make a better guess here since some consoles might not have support
1342 * for color mode for example.
32391275 1343 *
43db615b 1344 * However if TERM was configured through the kernel command line then leave it alone. */
1d84ad94 1345 r = proc_cmdline_get_key("TERM", 0, &term);
32391275
FB		 1346 if (r < 0)
1347 return r;
32391275 1348
4dc63c4b
LP		 1349 t = term ?: default_term_for_tty("/dev/console");
1350
1351 if (setenv("TERM", t, 1) < 0)
32391275
FB		 1352 return -errno;
1353
1354 return 0;
1355}
1356
6808a0bc
LP		 1357static void redirect_telinit(int argc, char *argv[]) {
1358
1359 /* This is compatibility support for SysV, where calling init as a user is identical to telinit. */
1360
1361#if HAVE_SYSV_COMPAT
1362 if (getpid_cached() == 1)
1363 return;
1364
1365 if (!strstr(program_invocation_short_name, "init"))
1366 return;
1367
1368 execv(SYSTEMCTL_BINARY_PATH, argv);
1369 log_error_errno(errno, "Failed to exec " SYSTEMCTL_BINARY_PATH ": %m");
a45d7127 1370 exit(EXIT_FAILURE);
6808a0bc
LP		 1371#endif
1372}
1373
4a36297c
LP		 1374static int become_shutdown(
1375 const char *shutdown_verb,
7eb35049 1376 int retval) {
4a36297c
LP		 1377
1378 char log_level[DECIMAL_STR_MAX(int) + 1],
e73c54b8
JK		 1379 exit_code[DECIMAL_STR_MAX(uint8_t) + 1],
1380 timeout[DECIMAL_STR_MAX(usec_t) + 1];
4a36297c 1381
e73c54b8 1382 const char* command_line[13] = {
4a36297c
LP		 1383 SYSTEMD_SHUTDOWN_BINARY_PATH,
1384 shutdown_verb,
e73c54b8 1385 "--timeout", timeout,
4a36297c
LP		 1386 "--log-level", log_level,
1387 "--log-target",
1388 };
1389
1390 _cleanup_strv_free_ char **env_block = NULL;
e73c54b8 1391 size_t pos = 7;
4a36297c
LP		 1392 int r;
1393
7eb35049 1394 assert(shutdown_verb);
234519ae 1395 assert(!command_line[pos]);
4a36297c
LP		 1396 env_block = strv_copy(environ);
1397
1398 xsprintf(log_level, "%d", log_get_max_level());
e73c54b8 1399 xsprintf(timeout, "%" PRI_USEC "us", arg_default_timeout_stop_usec);
4a36297c
LP		 1400
1401 switch (log_get_target()) {
1402
1403 case LOG_TARGET_KMSG:
1404 case LOG_TARGET_JOURNAL_OR_KMSG:
1405 case LOG_TARGET_SYSLOG_OR_KMSG:
1406 command_line[pos++] = "kmsg";
1407 break;
1408
1409 case LOG_TARGET_NULL:
1410 command_line[pos++] = "null";
1411 break;
1412
1413 case LOG_TARGET_CONSOLE:
1414 default:
1415 command_line[pos++] = "console";
1416 break;
1417 };
1418
1419 if (log_get_show_color())
1420 command_line[pos++] = "--log-color";
1421
1422 if (log_get_show_location())
1423 command_line[pos++] = "--log-location";
1424
1425 if (streq(shutdown_verb, "exit")) {
1426 command_line[pos++] = "--exit-code";
1427 command_line[pos++] = exit_code;
1428 xsprintf(exit_code, "%d", retval);
1429 }
1430
1431 assert(pos < ELEMENTSOF(command_line));
1432
7eb35049
LP		 1433 if (streq(shutdown_verb, "reboot") &&
1434 arg_shutdown_watchdog > 0 &&
1435 arg_shutdown_watchdog != USEC_INFINITY) {
1436
4a36297c
LP		 1437 char *e;
1438
1439 /* If we reboot let's set the shutdown
1440 * watchdog and tell the shutdown binary to
1441 * repeatedly ping it */
1442 r = watchdog_set_timeout(&arg_shutdown_watchdog);
1443 watchdog_close(r < 0);
1444
1445 /* Tell the binary how often to ping, ignore failure */
1446 if (asprintf(&e, "WATCHDOG_USEC="USEC_FMT, arg_shutdown_watchdog) > 0)
8a2c1fbf
EJ		 1447 (void) strv_consume(&env_block, e);
1448
1449 if (arg_watchdog_device &&
1450 asprintf(&e, "WATCHDOG_DEVICE=%s", arg_watchdog_device) > 0)
1451 (void) strv_consume(&env_block, e);
4a36297c
LP		 1452 } else
1453 watchdog_close(true);
1454
1455 /* Avoid the creation of new processes forked by the
1456 * kernel; at this point, we will not listen to the
1457 * signals anyway */
1458 if (detect_container() <= 0)
1459 (void) cg_uninstall_release_agent(SYSTEMD_CGROUP_CONTROLLER);
1460
1461 execve(SYSTEMD_SHUTDOWN_BINARY_PATH, (char **) command_line, env_block);
1462 return -errno;
1463}
1464
e839bafd
LP		 1465static void initialize_clock(void) {
1466 int r;
1467
1468 if (clock_is_localtime(NULL) > 0) {
1469 int min;
1470
1471 /*
1472 * The very first call of settimeofday() also does a time warp in the kernel.
1473 *
1474 * In the rtc-in-local time mode, we set the kernel's timezone, and rely on external tools to take care
1475 * of maintaining the RTC and do all adjustments. This matches the behavior of Windows, which leaves
1476 * the RTC alone if the registry tells that the RTC runs in UTC.
1477 */
1478 r = clock_set_timezone(&min);
1479 if (r < 0)
1480 log_error_errno(r, "Failed to apply local time delta, ignoring: %m");
1481 else
1482 log_info("RTC configured in localtime, applying delta of %i minutes to system time.", min);
1483
1484 } else if (!in_initrd()) {
1485 /*
1486 * Do a dummy very first call to seal the kernel's time warp magic.
1487 *
1488 * Do not call this from inside the initrd. The initrd might not carry /etc/adjtime with LOCAL, but the
1489 * real system could be set up that way. In such case, we need to delay the time-warp or the sealing
1490 * until we reach the real system.
1491 *
1492 * Do no set the kernel's timezone. The concept of local time cannot be supported reliably, the time
1493 * will jump or be incorrect at every daylight saving time change. All kernel local time concepts will
1494 * be treated as UTC that way.
1495 */
1496 (void) clock_reset_timewarp();
1497 }
1498
1499 r = clock_apply_epoch();
1500 if (r < 0)
1501 log_error_errno(r, "Current system time is before build time, but cannot correct: %m");
1502 else if (r > 0)
1503 log_info("System time before build time, advancing clock.");
1504}
1505
1e41242e 1506static void initialize_coredump(bool skip_setup) {
752bcb77 1507#if ENABLE_COREDUMP
1e41242e
LP		 1508 if (getpid_cached() != 1)
1509 return;
1510
1511 /* Don't limit the core dump size, so that coredump handlers such as systemd-coredump (which honour the limit)
1512 * will process core dumps for system services by default. */
1513 if (setrlimit(RLIMIT_CORE, &RLIMIT_MAKE_CONST(RLIM_INFINITY)) < 0)
1514 log_warning_errno(errno, "Failed to set RLIMIT_CORE: %m");
1515
c6885f5f
FB		 1516 /* But at the same time, turn off the core_pattern logic by default, so that no
1517 * coredumps are stored until the systemd-coredump tool is enabled via
1518 * sysctl. However it can be changed via the kernel command line later so core
1519 * dumps can still be generated during early startup and in initramfs. */
1e41242e 1520 if (!skip_setup)
e557b1a6 1521 disable_coredumps();
752bcb77 1522#endif
1e41242e
LP		 1523}
1524
c6885f5f
FB		 1525static void initialize_core_pattern(bool skip_setup) {
1526 int r;
1527
1528 if (skip_setup || !arg_early_core_pattern)
1529 return;
1530
1531 if (getpid_cached() != 1)
1532 return;
1533
1534 r = write_string_file("/proc/sys/kernel/core_pattern", arg_early_core_pattern, 0);
1535 if (r < 0)
1536 log_warning_errno(r, "Failed to write '%s' to /proc/sys/kernel/core_pattern, ignoring: %m", arg_early_core_pattern);
1537}
1538
3c7878f9
LP		 1539static void do_reexecute(
1540 int argc,
1541 char *argv[],
1542 const struct rlimit *saved_rlimit_nofile,
1543 const struct rlimit *saved_rlimit_memlock,
1544 FDSet *fds,
1545 const char *switch_root_dir,
1546 const char *switch_root_init,
1547 const char **ret_error_message) {
1548
1549 unsigned i, j, args_size;
1550 const char **args;
1551 int r;
1552
1553 assert(saved_rlimit_nofile);
1554 assert(saved_rlimit_memlock);
1555 assert(ret_error_message);
1556
1557 /* Close and disarm the watchdog, so that the new instance can reinitialize it, but doesn't get rebooted while
1558 * we do that */
1559 watchdog_close(true);
1560
1561 /* Reset the RLIMIT_NOFILE to the kernel default, so that the new systemd can pass the kernel default to its
1562 * child processes */
1563
1564 if (saved_rlimit_nofile->rlim_cur > 0)
1565 (void) setrlimit(RLIMIT_NOFILE, saved_rlimit_nofile);
1566 if (saved_rlimit_memlock->rlim_cur != (rlim_t) -1)
1567 (void) setrlimit(RLIMIT_MEMLOCK, saved_rlimit_memlock);
1568
1569 if (switch_root_dir) {
1570 /* Kill all remaining processes from the initrd, but don't wait for them, so that we can handle the
1571 * SIGCHLD for them after deserializing. */
e73c54b8 1572 broadcast_signal(SIGTERM, false, true, arg_default_timeout_stop_usec);
3c7878f9
LP		 1573
1574 /* And switch root with MS_MOVE, because we remove the old directory afterwards and detach it. */
1575 r = switch_root(switch_root_dir, "/mnt", true, MS_MOVE);
1576 if (r < 0)
1577 log_error_errno(r, "Failed to switch root, trying to continue: %m");
1578 }
1579
1580 args_size = MAX(6, argc+1);
1581 args = newa(const char*, args_size);
1582
1583 if (!switch_root_init) {
1584 char sfd[DECIMAL_STR_MAX(int) + 1];
1585
1586 /* First try to spawn ourselves with the right path, and with full serialization. We do this only if
1587 * the user didn't specify an explicit init to spawn. */
1588
1589 assert(arg_serialization);
1590 assert(fds);
1591
1592 xsprintf(sfd, "%i", fileno(arg_serialization));
1593
1594 i = 0;
1595 args[i++] = SYSTEMD_BINARY_PATH;
1596 if (switch_root_dir)
1597 args[i++] = "--switched-root";
1598 args[i++] = arg_system ? "--system" : "--user";
1599 args[i++] = "--deserialize";
1600 args[i++] = sfd;
1601 args[i++] = NULL;
1602
1603 assert(i <= args_size);
1604
1605 /*
1606 * We want valgrind to print its memory usage summary before reexecution. Valgrind won't do this is on
1607 * its own on exec(), but it will do it on exit(). Hence, to ensure we get a summary here, fork() off
1608 * a child, let it exit() cleanly, so that it prints the summary, and wait() for it in the parent,
1609 * before proceeding into the exec().
1610 */
1611 valgrind_summary_hack();
1612
1613 (void) execv(args[0], (char* const*) args);
1614 log_debug_errno(errno, "Failed to execute our own binary, trying fallback: %m");
1615 }
1616
1617 /* Try the fallback, if there is any, without any serialization. We pass the original argv[] and envp[]. (Well,
1618 * modulo the ordering changes due to getopt() in argv[], and some cleanups in envp[], but let's hope that
1619 * doesn't matter.) */
1620
1621 arg_serialization = safe_fclose(arg_serialization);
1622 fds = fdset_free(fds);
1623
1624 /* Reopen the console */
1625 (void) make_console_stdio();
1626
1627 for (j = 1, i = 1; j < (unsigned) argc; j++)
1628 args[i++] = argv[j];
1629 args[i++] = NULL;
1630 assert(i <= args_size);
1631
1632 /* Reenable any blocked signals, especially important if we switch from initial ramdisk to init=... */
1633 (void) reset_all_signal_handlers();
1634 (void) reset_signal_mask();
1635
1636 if (switch_root_init) {
1637 args[0] = switch_root_init;
1638 (void) execv(args[0], (char* const*) args);
1639 log_warning_errno(errno, "Failed to execute configured init, trying fallback: %m");
1640 }
1641
1642 args[0] = "/sbin/init";
1643 (void) execv(args[0], (char* const*) args);
1644 r = -errno;
1645
1646 manager_status_printf(NULL, STATUS_TYPE_EMERGENCY,
1647 ANSI_HIGHLIGHT_RED " !! " ANSI_NORMAL,
1648 "Failed to execute /sbin/init");
1649
1650 if (r == -ENOENT) {
1651 log_warning("No /sbin/init, trying fallback");
1652
1653 args[0] = "/bin/sh";
1654 args[1] = NULL;
1655 (void) execv(args[0], (char* const*) args);
1656 log_error_errno(errno, "Failed to execute /bin/sh, giving up: %m");
1657 } else
1658 log_warning_errno(r, "Failed to execute /sbin/init, giving up: %m");
1659
1660 *ret_error_message = "Failed to execute fallback shell";
1661}
1662
7eb35049
LP		 1663static int invoke_main_loop(
1664 Manager *m,
1665 bool *ret_reexecute,
1666 int *ret_retval, /* Return parameters relevant for shutting down */
1667 const char **ret_shutdown_verb, /* … */
1668 FDSet **ret_fds, /* Return parameters for reexecuting */
1669 char **ret_switch_root_dir, /* … */
1670 char **ret_switch_root_init, /* … */
1671 const char **ret_error_message) {
1672
1673 int r;
1674
1675 assert(m);
1676 assert(ret_reexecute);
1677 assert(ret_retval);
1678 assert(ret_shutdown_verb);
1679 assert(ret_fds);
1680 assert(ret_switch_root_dir);
1681 assert(ret_switch_root_init);
1682 assert(ret_error_message);
1683
1684 for (;;) {
1685 r = manager_loop(m);
1686 if (r < 0) {
1687 *ret_error_message = "Failed to run main loop";
1688 return log_emergency_errno(r, "Failed to run main loop: %m");
1689 }
1690
3ca4d0b3 1691 switch ((ManagerObjective) r) {
7eb35049 1692
a6ecbf83 1693 case MANAGER_RELOAD: {
bda7d78b 1694 LogTarget saved_log_target;
a6ecbf83
FB		 1695 int saved_log_level;
1696
7eb35049
LP		 1697 log_info("Reloading.");
1698
3fe91079 1699 /* First, save any overridden log level/target, then parse the configuration file, which might
bda7d78b
FB		 1700 * change the log level to new settings. */
1701
a6ecbf83 1702 saved_log_level = m->log_level_overridden ? log_get_max_level() : -1;
bda7d78b 1703 saved_log_target = m->log_target_overridden ? log_get_target() : _LOG_TARGET_INVALID;
a6ecbf83 1704
7eb35049
LP		 1705 r = parse_config_file();
1706 if (r < 0)
1707 log_warning_errno(r, "Failed to parse config file, ignoring: %m");
1708
1709 set_manager_defaults(m);
1710
a6ecbf83
FB		 1711 if (saved_log_level >= 0)
1712 manager_override_log_level(m, saved_log_level);
bda7d78b
FB		 1713 if (saved_log_target >= 0)
1714 manager_override_log_target(m, saved_log_target);
a6ecbf83 1715
7eb35049
LP		 1716 r = manager_reload(m);
1717 if (r < 0)
7a35fa24
LP		 1718 /* Reloading failed before the point of no return. Let's continue running as if nothing happened. */
1719 m->objective = MANAGER_OK;
7eb35049
LP		 1720
1721 break;
a6ecbf83 1722 }
7eb35049
LP		 1723
1724 case MANAGER_REEXECUTE:
1725
1726 r = prepare_reexecute(m, &arg_serialization, ret_fds, false);
1727 if (r < 0) {
1728 *ret_error_message = "Failed to prepare for reexecution";
1729 return r;
1730 }
1731
1732 log_notice("Reexecuting.");
1733
1734 *ret_reexecute = true;
1735 *ret_retval = EXIT_SUCCESS;
1736 *ret_shutdown_verb = NULL;
1737 *ret_switch_root_dir = *ret_switch_root_init = NULL;
1738
1739 return 0;
1740
1741 case MANAGER_SWITCH_ROOT:
1742 if (!m->switch_root_init) {
1743 r = prepare_reexecute(m, &arg_serialization, ret_fds, true);
1744 if (r < 0) {
1745 *ret_error_message = "Failed to prepare for reexecution";
1746 return r;
1747 }
1748 } else
1749 *ret_fds = NULL;
1750
1751 log_notice("Switching root.");
1752
1753 *ret_reexecute = true;
1754 *ret_retval = EXIT_SUCCESS;
1755 *ret_shutdown_verb = NULL;
1756
1757 /* Steal the switch root parameters */
1758 *ret_switch_root_dir = m->switch_root;
1759 *ret_switch_root_init = m->switch_root_init;
1760 m->switch_root = m->switch_root_init = NULL;
1761
1762 return 0;
1763
1764 case MANAGER_EXIT:
1765
1766 if (MANAGER_IS_USER(m)) {
1767 log_debug("Exit.");
1768
1769 *ret_reexecute = false;
1770 *ret_retval = m->return_value;
1771 *ret_shutdown_verb = NULL;
1772 *ret_fds = NULL;
1773 *ret_switch_root_dir = *ret_switch_root_init = NULL;
1774
1775 return 0;
1776 }
1777
1778 _fallthrough_;
1779 case MANAGER_REBOOT:
1780 case MANAGER_POWEROFF:
1781 case MANAGER_HALT:
1782 case MANAGER_KEXEC: {
af41e508
LP		 1783 static const char * const table[_MANAGER_OBJECTIVE_MAX] = {
1784 [MANAGER_EXIT] = "exit",
1785 [MANAGER_REBOOT] = "reboot",
7eb35049 1786 [MANAGER_POWEROFF] = "poweroff",
af41e508
LP		 1787 [MANAGER_HALT] = "halt",
1788 [MANAGER_KEXEC] = "kexec",
7eb35049
LP		 1789 };
1790
1791 log_notice("Shutting down.");
1792
1793 *ret_reexecute = false;
1794 *ret_retval = m->return_value;
af41e508 1795 assert_se(*ret_shutdown_verb = table[m->objective]);
7eb35049
LP		 1796 *ret_fds = NULL;
1797 *ret_switch_root_dir = *ret_switch_root_init = NULL;
1798
1799 return 0;
1800 }
1801
1802 default:
af41e508 1803 assert_not_reached("Unknown or unexpected manager objective.");
7eb35049
LP		 1804 }
1805 }
1806}
1807
31aef7ff
LP		 1808static void log_execution_mode(bool *ret_first_boot) {
1809 assert(ret_first_boot);
1810
1811 if (arg_system) {
1812 int v;
1813
1814 log_info(PACKAGE_STRING " running in %ssystem mode. (" SYSTEMD_FEATURES ")",
1815 arg_action == ACTION_TEST ? "test " : "" );
1816
1817 v = detect_virtualization();
1818 if (v > 0)
1819 log_info("Detected virtualization %s.", virtualization_to_string(v));
1820
1821 log_info("Detected architecture %s.", architecture_to_string(uname_architecture()));
1822
1823 if (in_initrd()) {
1824 *ret_first_boot = false;
1825 log_info("Running in initial RAM disk.");
1826 } else {
1827 /* Let's check whether we are in first boot, i.e. whether /etc is still unpopulated. We use
1828 * /etc/machine-id as flag file, for this: if it exists we assume /etc is populated, if it
1829 * doesn't it's unpopulated. This allows container managers and installers to provision a
1830 * couple of files already. If the container manager wants to provision the machine ID itself
1831 * it should pass $container_uuid to PID 1. */
1832
1833 *ret_first_boot = access("/etc/machine-id", F_OK) < 0;
1834 if (*ret_first_boot)
1835 log_info("Running with unpopulated /etc.");
1836 }
1837 } else {
b9e90f3a
LP		 1838 if (DEBUG_LOGGING) {
1839 _cleanup_free_ char *t;
31aef7ff 1840
b9e90f3a
LP		 1841 t = uid_to_name(getuid());
1842 log_debug(PACKAGE_STRING " running in %suser mode for user " UID_FMT "/%s. (" SYSTEMD_FEATURES ")",
1843 arg_action == ACTION_TEST ? " test" : "", getuid(), strna(t));
1844 }
31aef7ff
LP		 1845
1846 *ret_first_boot = false;
1847 }
1848}
1849
5afbaa36
LP		 1850static int initialize_runtime(
1851 bool skip_setup,
1852 struct rlimit *saved_rlimit_nofile,
1853 struct rlimit *saved_rlimit_memlock,
1854 const char **ret_error_message) {
1855
1856 int r;
1857
1858 assert(ret_error_message);
1859
1860 /* Sets up various runtime parameters. Many of these initializations are conditionalized:
1861 *
1862 * - Some only apply to --system instances
1863 * - Some only apply to --user instances
1864 * - Some only apply when we first start up, but not when we reexecute
1865 */
1866
2d776038
LP		 1867 if (arg_action != ACTION_RUN)
1868 return 0;
1869
3c3c6cb9
LP		 1870 if (arg_system) {
1871 /* Make sure we leave a core dump without panicing the kernel. */
1872 install_crash_handler();
5afbaa36 1873
3c3c6cb9
LP		 1874 if (!skip_setup) {
1875 r = mount_cgroup_controllers(arg_join_controllers);
1876 if (r < 0) {
1877 *ret_error_message = "Failed to mount cgroup hierarchies";
1878 return r;
1879 }
1880
1881 status_welcome();
1882 hostname_setup();
1883 machine_id_setup(NULL, arg_machine_id, NULL);
1884 loopback_setup();
1885 bump_unix_max_dgram_qlen();
1886 test_usr();
1887 write_container_id();
1888 }
8a2c1fbf 1889
3c3c6cb9
LP		 1890 if (arg_watchdog_device) {
1891 r = watchdog_set_device(arg_watchdog_device);
1892 if (r < 0)
1893 log_warning_errno(r, "Failed to set watchdog device to %s, ignoring: %m", arg_watchdog_device);
1894 }
1895
1896 if (arg_runtime_watchdog > 0 && arg_runtime_watchdog != USEC_INFINITY)
1897 watchdog_set_timeout(&arg_runtime_watchdog);
1898 }
5afbaa36
LP		 1899
1900 if (arg_timer_slack_nsec != NSEC_INFINITY)
1901 if (prctl(PR_SET_TIMERSLACK, arg_timer_slack_nsec) < 0)
3a671cd1 1902 log_warning_errno(errno, "Failed to adjust timer slack, ignoring: %m");
5afbaa36
LP		 1903
1904 if (arg_system && !cap_test_all(arg_capability_bounding_set)) {
1905 r = capability_bounding_set_drop_usermode(arg_capability_bounding_set);
1906 if (r < 0) {
1907 *ret_error_message = "Failed to drop capability bounding set of usermode helpers";
1908 return log_emergency_errno(r, "Failed to drop capability bounding set of usermode helpers: %m");
1909 }
1910
1911 r = capability_bounding_set_drop(arg_capability_bounding_set, true);
1912 if (r < 0) {
1913 *ret_error_message = "Failed to drop capability bounding set";
1914 return log_emergency_errno(r, "Failed to drop capability bounding set: %m");
1915 }
1916 }
1917
39362f6f
JB		 1918 if (arg_system && arg_no_new_privs) {
1919 if (prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0) < 0) {
1920 *ret_error_message = "Failed to disable new privileges";
1921 return log_emergency_errno(errno, "Failed to disable new privileges: %m");
1922 }
1923 }
1924
5afbaa36
LP		 1925 if (arg_syscall_archs) {
1926 r = enforce_syscall_archs(arg_syscall_archs);
1927 if (r < 0) {
1928 *ret_error_message = "Failed to set syscall architectures";
1929 return r;
1930 }
1931 }
1932
1933 if (!arg_system)
1934 /* Become reaper of our children */
1935 if (prctl(PR_SET_CHILD_SUBREAPER, 1) < 0)
1936 log_warning_errno(errno, "Failed to make us a subreaper: %m");
1937
a17c1712
LP		 1938 /* Bump up RLIMIT_NOFILE for systemd itself */
1939 (void) bump_rlimit_nofile(saved_rlimit_nofile);
1940 (void) bump_rlimit_memlock(saved_rlimit_memlock);
5afbaa36
LP		 1941
1942 return 0;
1943}
1944
6acca5fc
LP		 1945static int do_queue_default_job(
1946 Manager *m,
1947 const char **ret_error_message) {
1948
1949 _cleanup_(sd_bus_error_free) sd_bus_error error = SD_BUS_ERROR_NULL;
1950 Job *default_unit_job;
1951 Unit *target = NULL;
1952 int r;
1953
1954 log_debug("Activating default unit: %s", arg_default_unit);
1955
4109ede7
ZJS		 1956 r = manager_load_startable_unit_or_warn(m, arg_default_unit, NULL, &target);
1957 if (r < 0) {
1958 log_info("Falling back to rescue target: " SPECIAL_RESCUE_TARGET);
6acca5fc 1959
4109ede7 1960 r = manager_load_startable_unit_or_warn(m, SPECIAL_RESCUE_TARGET, NULL, &target);
6acca5fc 1961 if (r < 0) {
4109ede7
ZJS		 1962 *ret_error_message = r == -ERFKILL ? "Rescue target masked"
1963 : "Failed to load rescue target";
1964 return r;
6acca5fc
LP		 1965 }
1966 }
1967
1968 assert(target->load_state == UNIT_LOADED);
1969
1970 r = manager_add_job(m, JOB_START, target, JOB_ISOLATE, &error, &default_unit_job);
1971 if (r == -EPERM) {
1972 log_debug_errno(r, "Default target could not be isolated, starting instead: %s", bus_error_message(&error, r));
1973
1974 sd_bus_error_free(&error);
1975
1976 r = manager_add_job(m, JOB_START, target, JOB_REPLACE, &error, &default_unit_job);
1977 if (r < 0) {
1978 *ret_error_message = "Failed to start default target";
1979 return log_emergency_errno(r, "Failed to start default target: %s", bus_error_message(&error, r));
1980 }
1981
1982 } else if (r < 0) {
1983 *ret_error_message = "Failed to isolate default target";
1984 return log_emergency_errno(r, "Failed to isolate default target: %s", bus_error_message(&error, r));
1985 }
1986
1987 m->default_unit_job_id = default_unit_job->id;
1988
1989 return 0;
1990}
1991
970777b5 1992static void free_arguments(void) {
970777b5
LP		 1993
1994 /* Frees all arg_* variables, with the exception of arg_serialization */
ef59409f 1995 rlimit_free_all(arg_default_rlimit);
970777b5
LP		 1996
1997 arg_default_unit = mfree(arg_default_unit);
1998 arg_confirm_spawn = mfree(arg_confirm_spawn);
1999 arg_join_controllers = strv_free_free(arg_join_controllers);
2000 arg_default_environment = strv_free(arg_default_environment);
2001 arg_syscall_archs = set_free(arg_syscall_archs);
2002}
2003
97d1fb94
LP		 2004static int load_configuration(int argc, char **argv, const char **ret_error_message) {
2005 int r;
2006
2007 assert(ret_error_message);
2008
2009 arg_default_tasks_max = system_tasks_max_scale(DEFAULT_TASKS_MAX_PERCENTAGE, 100U);
2010
2011 r = parse_config_file();
2012 if (r < 0) {
2013 *ret_error_message = "Failed to parse config file";
2014 return r;
2015 }
2016
2017 if (arg_system) {
2018 r = proc_cmdline_parse(parse_proc_cmdline_item, NULL, 0);
2019 if (r < 0)
2020 log_warning_errno(r, "Failed to parse kernel command line, ignoring: %m");
2021 }
2022
2023 /* Note that this also parses bits from the kernel command line, including "debug". */
2024 log_parse_environment();
2025
2026 r = parse_argv(argc, argv);
2027 if (r < 0) {
2028 *ret_error_message = "Failed to parse commandline arguments";
2029 return r;
2030 }
2031
2032 /* Initialize default unit */
2033 if (!arg_default_unit) {
2034 arg_default_unit = strdup(SPECIAL_DEFAULT_TARGET);
2035 if (!arg_default_unit) {
2036 *ret_error_message = "Failed to set default unit";
2037 return log_oom();
2038 }
2039 }
2040
db33214b 2041 /* Initialize the show status setting if it hasn't been set explicitly yet */
7a293242 2042 if (arg_show_status == _SHOW_STATUS_INVALID)
db33214b
LP		 2043 arg_show_status = SHOW_STATUS_YES;
2044
97d1fb94
LP		 2045 return 0;
2046}
2047
b0d7c989
LP		 2048static int safety_checks(void) {
2049
febf46a4
LP		 2050 if (getpid_cached() == 1 &&
2051 arg_action != ACTION_RUN) {
2052 log_error("Unsupported execution mode while PID 1.");
2053 return -EPERM;
2054 }
2055
2056 if (getpid_cached() == 1 &&
2057 !arg_system) {
2058 log_error("Can't run --user mode as PID 1.");
2059 return -EPERM;
2060 }
2061
2062 if (arg_action == ACTION_RUN &&
2063 arg_system &&
2064 getpid_cached() != 1) {
2065 log_error("Can't run system mode unless PID 1.");
2066 return -EPERM;
2067 }
2068
b0d7c989
LP		 2069 if (arg_action == ACTION_TEST &&
2070 geteuid() == 0) {
2071 log_error("Don't run test mode as root.");
2072 return -EPERM;
2073 }
2074
2075 if (!arg_system &&
2076 arg_action == ACTION_RUN &&
2077 sd_booted() <= 0) {
2078 log_error("Trying to run as user instance, but the system has not been booted with systemd.");
2079 return -EOPNOTSUPP;
2080 }
2081
2082 if (!arg_system &&
2083 arg_action == ACTION_RUN &&
2084 !getenv("XDG_RUNTIME_DIR")) {
2085 log_error("Trying to run as user instance, but $XDG_RUNTIME_DIR is not set.");
2086 return -EUNATCH;
2087 }
2088
2089 if (arg_system &&
2090 arg_action == ACTION_RUN &&
2091 running_in_chroot() > 0) {
2092 log_error("Cannot be run in a chroot() environment.");
2093 return -EOPNOTSUPP;
2094 }
2095
2096 return 0;
2097}
2098
74da609f
LP		 2099static int initialize_security(
2100 bool *loaded_policy,
2101 dual_timestamp *security_start_timestamp,
2102 dual_timestamp *security_finish_timestamp,
2103 const char **ret_error_message) {
2104
2105 int r;
2106
2107 assert(loaded_policy);
2108 assert(security_start_timestamp);
2109 assert(security_finish_timestamp);
2110 assert(ret_error_message);
2111
2112 dual_timestamp_get(security_start_timestamp);
2113
97149f40 2114 r = mac_selinux_setup(loaded_policy);
74da609f
LP		 2115 if (r < 0) {
2116 *ret_error_message = "Failed to load SELinux policy";
2117 return r;
2118 }
2119
2120 r = mac_smack_setup(loaded_policy);
2121 if (r < 0) {
2122 *ret_error_message = "Failed to load SMACK policy";
2123 return r;
2124 }
2125
2126 r = ima_setup();
2127 if (r < 0) {
2128 *ret_error_message = "Failed to load IMA policy";
2129 return r;
2130 }
2131
2132 dual_timestamp_get(security_finish_timestamp);
2133 return 0;
2134}
2135
263162da
LP		 2136static void test_summary(Manager *m) {
2137 assert(m);
2138
2139 printf("-> By units:\n");
2140 manager_dump_units(m, stdout, "\t");
2141
2142 printf("-> By jobs:\n");
2143 manager_dump_jobs(m, stdout, "\t");
2144}
2145
efeb853f
LP		 2146static int collect_fds(FDSet **ret_fds, const char **ret_error_message) {
2147 int r;
2148
2149 assert(ret_fds);
2150 assert(ret_error_message);
2151
2152 r = fdset_new_fill(ret_fds);
2153 if (r < 0) {
2154 *ret_error_message = "Failed to allocate fd set";
2155 return log_emergency_errno(r, "Failed to allocate fd set: %m");
2156 }
2157
2158 fdset_cloexec(*ret_fds, true);
2159
2160 if (arg_serialization)
2161 assert_se(fdset_remove(*ret_fds, fileno(arg_serialization)) >= 0);
2162
2163 return 0;
2164}
2165
2e51b31c
LP		 2166static void setup_console_terminal(bool skip_setup) {
2167
2168 if (!arg_system)
2169 return;
2170
2171 /* Become a session leader if we aren't one yet. */
2172 (void) setsid();
2173
2174 /* If we are init, we connect stdin/stdout/stderr to /dev/null and make sure we don't have a controlling
2175 * tty. */
2176 (void) release_terminal();
2177
2178 /* Reset the console, but only if this is really init and we are freshly booted */
2179 if (getpid_cached() == 1 && !skip_setup)
2180 (void) console_setup();
2181}
2182
aa40ff07
LP		 2183static bool early_skip_setup_check(int argc, char *argv[]) {
2184 bool found_deserialize = false;
2185 int i;
2186
2187 /* Determine if this is a reexecution or normal bootup. We do the full command line parsing much later, so
2188 * let's just have a quick peek here. Note that if we have switched root, do all the special setup things
2189 * anyway, even if in that case we also do deserialization. */
2190
2191 for (i = 1; i < argc; i++) {
aa40ff07
LP		 2192 if (streq(argv[i], "--switched-root"))
2193 return false; /* If we switched root, don't skip the setup. */
2194 else if (streq(argv[i], "--deserialize"))
2195 found_deserialize = true;
2196 }
2197
2198 return found_deserialize; /* When we are deserializing, then we are reexecuting, hence avoid the extensive setup */
2199}
2200
60918275 2201int main(int argc, char *argv[]) {
625e8690
LP		 2202
2203 dual_timestamp initrd_timestamp = DUAL_TIMESTAMP_NULL, userspace_timestamp = DUAL_TIMESTAMP_NULL, kernel_timestamp = DUAL_TIMESTAMP_NULL,
2204 security_start_timestamp = DUAL_TIMESTAMP_NULL, security_finish_timestamp = DUAL_TIMESTAMP_NULL;
2205 struct rlimit saved_rlimit_nofile = RLIMIT_MAKE_CONST(0), saved_rlimit_memlock = RLIMIT_MAKE_CONST((rlim_t) -1);
2206 bool skip_setup, loaded_policy = false, queue_default_job = false, first_boot = false, reexecute = false;
2207 char *switch_root_dir = NULL, *switch_root_init = NULL;
9d76d730 2208 usec_t before_startup, after_startup;
625e8690 2209 static char systemd[] = "systemd";
9d76d730 2210 char timespan[FORMAT_TIMESPAN_MAX];
625e8690
LP		 2211 const char *shutdown_verb = NULL, *error_message = NULL;
2212 int r, retval = EXIT_FAILURE;
2213 Manager *m = NULL;
a16e1123 2214 FDSet *fds = NULL;
27b14a22 2215
d72a8f10 2216 /* SysV compatibility: redirect init → telinit */
6808a0bc 2217 redirect_telinit(argc, argv);
2cb1a60d 2218
d72a8f10 2219 /* Take timestamps early on */
c3a170f3
HH		 2220 dual_timestamp_from_monotonic(&kernel_timestamp, 0);
2221 dual_timestamp_get(&userspace_timestamp);
2222
d72a8f10
LP		 2223 /* Figure out whether we need to do initialize the system, or if we already did that because we are
2224 * reexecuting */
aa40ff07 2225 skip_setup = early_skip_setup_check(argc, argv);
d03bc1b8 2226
d72a8f10
LP		 2227 /* If we get started via the /sbin/init symlink then we are called 'init'. After a subsequent reexecution we
2228 * are then called 'systemd'. That is confusing, hence let's call us systemd right-away. */
f3b6a3ed 2229 program_invocation_short_name = systemd;
eee8b7ab 2230 (void) prctl(PR_SET_NAME, systemd);
5d6b1584 2231
d72a8f10 2232 /* Save the original command line */
9a0e6896
LP		 2233 saved_argv = argv;
2234 saved_argc = argc;
f3b6a3ed 2235
6fdb8de4 2236 /* Make sure that if the user says "syslog" we actually log to the journal. */
c1dc6153 2237 log_set_upgrade_syslog_to_journal(true);
bbe63281 2238
df0ff127 2239 if (getpid_cached() == 1) {
b5752d23
LP		 2240 /* When we run as PID 1 force system mode */
2241 arg_system = true;
2242
48a601fe 2243 /* Disable the umask logic */
90dc8c2e
MG		 2244 umask(0);
2245
92890452
LP		 2246 /* Make sure that at least initially we do not ever log to journald/syslogd, because it might not be
2247 * activated yet (even though the log socket for it exists). */
d075092f
LP		 2248 log_set_prohibit_ipc(true);
2249
48a601fe
LP		 2250 /* Always reopen /dev/console when running as PID 1 or one of its pre-execve() children. This is
2251 * important so that we never end up logging to any foreign stderr, for example if we have to log in a
2252 * child process right before execve()'ing the actual binary, at a point in time where socket
2253 * activation stderr/stdout area already set up. */
2254 log_set_always_reopen_console(true);
48a601fe 2255
92890452 2256 if (detect_container() <= 0) {
4f8d551f 2257
92890452 2258 /* Running outside of a container as PID 1 */
92890452
LP		 2259 log_set_target(LOG_TARGET_KMSG);
2260 log_open();
a866073d 2261
92890452
LP		 2262 if (in_initrd())
2263 initrd_timestamp = userspace_timestamp;
c3ba6250 2264
92890452
LP		 2265 if (!skip_setup) {
2266 r = mount_setup_early();
2267 if (r < 0) {
2268 error_message = "Failed to mount early API filesystems";
2269 goto finish;
2270 }
2271
2272 r = initialize_security(
2273 &loaded_policy,
2274 &security_start_timestamp,
2275 &security_finish_timestamp,
2276 &error_message);
2277 if (r < 0)
2278 goto finish;
d723cd65 2279 }
eee8b7ab 2280
92890452
LP		 2281 if (mac_selinux_init() < 0) {
2282 error_message = "Failed to initialize SELinux policy";
96694e99 2283 goto finish;
92890452 2284 }
0b3325e7 2285
92890452
LP		 2286 if (!skip_setup)
2287 initialize_clock();
2288
2289 /* Set the default for later on, but don't actually open the logs like this for now. Note that
2290 * if we are transitioning from the initrd there might still be journal fd open, and we
2291 * shouldn't attempt opening that before we parsed /proc/cmdline which might redirect output
2292 * elsewhere. */
2293 log_set_target(LOG_TARGET_JOURNAL_OR_KMSG);
2294
2295 } else {
2296 /* Running inside a container, as PID 1 */
92890452
LP		 2297 log_set_target(LOG_TARGET_CONSOLE);
2298 log_open();
2299
2300 /* For later on, see above... */
2301 log_set_target(LOG_TARGET_JOURNAL);
2302
2303 /* clear the kernel timestamp,
2304 * because we are in a container */
2305 kernel_timestamp = DUAL_TIMESTAMP_NULL;
cb6531be 2306 }
7948c4df 2307
92890452 2308 initialize_coredump(skip_setup);
a866073d 2309
92890452
LP		 2310 r = fixup_environment();
2311 if (r < 0) {
2312 log_emergency_errno(r, "Failed to fix up PID 1 environment: %m");
2313 error_message = "Failed to fix up PID1 environment";
2314 goto finish;
2315 }
a866073d 2316
c3a170f3 2317 } else {
a866073d 2318 /* Running as user instance */
463d0d15 2319 arg_system = false;
eeecf6e6 2320 log_set_target(LOG_TARGET_AUTO);
871e5809 2321 log_open();
c3a170f3
HH		 2322
2323 /* clear the kernel timestamp,
2324 * because we are not PID 1 */
6513d561 2325 kernel_timestamp = DUAL_TIMESTAMP_NULL;
bbe63281 2326 }
a5dab5ce 2327
6edefe0b 2328 if (arg_system) {
92890452
LP		 2329 /* Try to figure out if we can use colors with the console. No need to do that for user instances since
2330 * they never log into the console. */
3a18b604 2331 log_show_color(colors_enabled());
92890452 2332
c76cf844
AK		 2333 r = make_null_stdio();
2334 if (r < 0)
92890452 2335 log_warning_errno(r, "Failed to redirect standard streams to /dev/null, ignoring: %m");
3a18b604
FB		 2336 }
2337
f170852a
LP		 2338 /* Mount /proc, /sys and friends, so that /proc/cmdline and
2339 * /proc/$PID/fd is available. */
df0ff127 2340 if (getpid_cached() == 1) {
f84f9974 2341
a132bef0 2342 /* Load the kernel modules early. */
2e75e2a8
DM		 2343 if (!skip_setup)
2344 kmod_setup();
2e75e2a8 2345
0c85a4f3 2346 r = mount_setup(loaded_policy);
cb6531be
ZJS		 2347 if (r < 0) {
2348 error_message = "Failed to mount API filesystems";
8efe3c01 2349 goto finish;
cb6531be 2350 }
0c85a4f3 2351 }
4ade7963
LP		 2352
2353 /* Reset all signal handlers. */
ce30c8dc
LP		 2354 (void) reset_all_signal_handlers();
2355 (void) ignore_signals(SIGNALS_IGNORE, -1);
078e4539 2356
97d1fb94
LP		 2357 r = load_configuration(argc, argv, &error_message);
2358 if (r < 0)
f170852a 2359 goto finish;
10c961b9 2360
b0d7c989
LP		 2361 r = safety_checks();
2362 if (r < 0)
fe783b03 2363 goto finish;
fe783b03 2364
bbc1acab 2365 if (IN_SET(arg_action, ACTION_TEST, ACTION_HELP, ACTION_DUMP_CONFIGURATION_ITEMS, ACTION_DUMP_BUS_PROPERTIES))
ee5324aa 2366 (void) pager_open(arg_no_pager, false);
b0d7c989
LP		 2367
2368 if (arg_action != ACTION_RUN)
74e7579c 2369 skip_setup = true;
b87c2aa6 2370
fa0f4d8a 2371 if (arg_action == ACTION_HELP) {
37ec0fdd 2372 retval = help() < 0 ? EXIT_FAILURE : EXIT_SUCCESS;
f170852a 2373 goto finish;
9ba0bc4e
ZJS		 2374 } else if (arg_action == ACTION_VERSION) {
2375 retval = version();
2376 goto finish;
fa0f4d8a 2377 } else if (arg_action == ACTION_DUMP_CONFIGURATION_ITEMS) {
e537352b 2378 unit_dump_config_items(stdout);
22f4096c 2379 retval = EXIT_SUCCESS;
e537352b 2380 goto finish;
bbc1acab
YW		 2381 } else if (arg_action == ACTION_DUMP_BUS_PROPERTIES) {
2382 dump_bus_properties(stdout);
2383 retval = EXIT_SUCCESS;
2384 goto finish;
f170852a
LP		 2385 }
2386
4c701096 2387 assert_se(IN_SET(arg_action, ACTION_RUN, ACTION_TEST));
f170852a 2388
5a2e0c62
LP		 2389 /* Move out of the way, so that we won't block unmounts */
2390 assert_se(chdir("/") == 0);
2391
dea374e8 2392 if (arg_action == ACTION_RUN) {
a70c72a0 2393
c6885f5f
FB		 2394 /* A core pattern might have been specified via the cmdline. */
2395 initialize_core_pattern(skip_setup);
2396
efeb853f 2397 /* Close logging fds, in order not to confuse collecting passed fds and terminal logic below */
a70c72a0
LP		 2398 log_close();
2399
2400 /* Remember open file descriptors for later deserialization */
efeb853f
LP		 2401 r = collect_fds(&fds, &error_message);
2402 if (r < 0)
dea374e8 2403 goto finish;
a16e1123 2404
2e51b31c
LP		 2405 /* Give up any control of the console, but make sure its initialized. */
2406 setup_console_terminal(skip_setup);
56d96fc0 2407
a70c72a0
LP		 2408 /* Open the logging devices, if possible and necessary */
2409 log_open();
56d96fc0 2410 }
4ade7963 2411
31aef7ff 2412 log_execution_mode(&first_boot);
a5dab5ce 2413
2d776038
LP		 2414 r = initialize_runtime(skip_setup,
2415 &saved_rlimit_nofile,
2416 &saved_rlimit_memlock,
2417 &error_message);
2418 if (r < 0)
2419 goto finish;
4096d6f5 2420
e0a3da1f
ZJS		 2421 r = manager_new(arg_system ? UNIT_FILE_SYSTEM : UNIT_FILE_USER,
2422 arg_action == ACTION_TEST ? MANAGER_TEST_FULL : 0,
2423 &m);
e96d6be7 2424 if (r < 0) {
da927ba9 2425 log_emergency_errno(r, "Failed to allocate manager object: %m");
cb6531be 2426 error_message = "Failed to allocate manager object";
60918275
LP		 2427 goto finish;
2428 }
2429
9f9f0342
LP		 2430 m->timestamps[MANAGER_TIMESTAMP_KERNEL] = kernel_timestamp;
2431 m->timestamps[MANAGER_TIMESTAMP_INITRD] = initrd_timestamp;
2432 m->timestamps[MANAGER_TIMESTAMP_USERSPACE] = userspace_timestamp;
d4ee7bd8
YW		 2433 m->timestamps[manager_timestamp_initrd_mangle(MANAGER_TIMESTAMP_SECURITY_START)] = security_start_timestamp;
2434 m->timestamps[manager_timestamp_initrd_mangle(MANAGER_TIMESTAMP_SECURITY_FINISH)] = security_finish_timestamp;
9e58ff9c 2435
85cb4151 2436 set_manager_defaults(m);
7b46fc6a 2437 set_manager_settings(m);
fd130612 2438 manager_set_first_boot(m, first_boot);
27d340c7 2439
bf4df7c3 2440 /* Remember whether we should queue the default job */
d3b1c508 2441 queue_default_job = !arg_serialization || arg_switched_root;
bf4df7c3 2442
9d76d730
LP		 2443 before_startup = now(CLOCK_MONOTONIC);
2444
d3b1c508 2445 r = manager_startup(m, arg_serialization, fds);
58f88d92 2446 if (r < 0) {
cefb3eda 2447 error_message = "Failed to start up manager";
58f88d92
ZJS		 2448 goto finish;
2449 }
a16e1123 2450
6acca5fc 2451 /* This will close all file descriptors that were opened, but not claimed by any unit. */
2feceb5e 2452 fds = fdset_free(fds);
74ca738f 2453 arg_serialization = safe_fclose(arg_serialization);
bf4df7c3
LP		 2454
2455 if (queue_default_job) {
6acca5fc 2456 r = do_queue_default_job(m, &error_message);
718db961 2457 if (r < 0)
37d88da7 2458 goto finish;
6acca5fc 2459 }
ab17a050 2460
6acca5fc 2461 after_startup = now(CLOCK_MONOTONIC);
60918275 2462
6acca5fc
LP		 2463 log_full(arg_action == ACTION_TEST ? LOG_INFO : LOG_DEBUG,
2464 "Loaded units and determined initial transaction in %s.",
2465 format_timespan(timespan, sizeof(timespan), after_startup - before_startup, 100 * USEC_PER_MSEC));
07672f49 2466
6acca5fc 2467 if (arg_action == ACTION_TEST) {
263162da 2468 test_summary(m);
6acca5fc
LP		 2469 retval = EXIT_SUCCESS;
2470 goto finish;
e965d56d 2471 }
d46de8a1 2472
3046b6db
LP		 2473 (void) invoke_main_loop(m,
2474 &reexecute,
2475 &retval,
2476 &shutdown_verb,
2477 &fds,
2478 &switch_root_dir,
2479 &switch_root_init,
2480 &error_message);
f170852a 2481
60918275 2482finish:
b87c2aa6
ZJS		 2483 pager_close();
2484
92890452 2485 if (m) {
81f5fc2d 2486 arg_shutdown_watchdog = m->shutdown_watchdog;
92890452
LP		 2487 m = manager_free(m);
2488 }
60918275 2489
970777b5 2490 free_arguments();
cc56fafe 2491 mac_selinux_finish();
b2bb3dbe 2492
3c7878f9
LP		 2493 if (reexecute)
2494 do_reexecute(argc, argv,
2495 &saved_rlimit_nofile,
2496 &saved_rlimit_memlock,
2497 fds,
2498 switch_root_dir,
2499 switch_root_init,
2500 &error_message); /* This only returns if reexecution failed */
a16e1123 2501
74ca738f 2502 arg_serialization = safe_fclose(arg_serialization);
2feceb5e 2503 fds = fdset_free(fds);
a16e1123 2504
349cc4a5 2505#if HAVE_VALGRIND_VALGRIND_H
54b434b1
LP		 2506 /* If we are PID 1 and running under valgrind, then let's exit
2507 * here explicitly. valgrind will only generate nice output on
2508 * exit(), not on exec(), hence let's do the former not the
2509 * latter here. */
8a2c1fbf
EJ		 2510 if (getpid_cached() == 1 && RUNNING_ON_VALGRIND) {
2511 /* Cleanup watchdog_device strings for valgrind. We need them
2512 * in become_shutdown() so normally we cannot free them yet. */
2513 watchdog_free_device();
2514 arg_watchdog_device = mfree(arg_watchdog_device);
27fe58b7 2515 return retval;
8a2c1fbf 2516 }
54b434b1
LP		 2517#endif
2518
b9080b03 2519 if (shutdown_verb) {
7eb35049 2520 r = become_shutdown(shutdown_verb, retval);
4a36297c 2521 log_error_errno(r, "Failed to execute shutdown binary, %s: %m", getpid_cached() == 1 ? "freezing" : "quitting");
9b9881d7 2522 error_message = "Failed to execute shutdown binary";
b9080b03
FF		 2523 }
2524
8a2c1fbf
EJ		 2525 watchdog_free_device();
2526 arg_watchdog_device = mfree(arg_watchdog_device);
2527
df0ff127 2528 if (getpid_cached() == 1) {
cb6531be
ZJS		 2529 if (error_message)
2530 manager_status_printf(NULL, STATUS_TYPE_EMERGENCY,
1fc464f6 2531 ANSI_HIGHLIGHT_RED "!!!!!!" ANSI_NORMAL,
cb6531be 2532 "%s, freezing.", error_message);
b9e74c39 2533 freeze_or_reboot();
cb6531be 2534 }
c3b3c274 2535
60918275
LP		 2536 return retval;
2537}
Unnamed repository; edit this file 'description' to name the repository.